JP2018536364A - Multi-region search range for block prediction mode for display stream compression (DSC) - Google Patents

Abstract

  A method for coding a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme for transmission over a display link is disclosed. In one aspect, the method includes determining a candidate block that is used to predict a current block in a current slice, wherein the candidate block is a plurality of pixels each corresponding to a reconstructed pixel in the current slice. Within range. The range of pixel locations includes: (i) a first region that includes one or more first pixel locations in a first line of a plurality of pixels that overlaps the current block; and (ii) an overlap with the current block. And a second region including one or more second pixel locations in a second line of non-wrapping pixels. The method may further comprise determining and signaling a prediction vector that indicates a pixel location of the candidate block.

Description

  [0001] The present disclosure relates to the field of video coding and compression, and in particular, to video compression for transmission over a display link, such as display link video compression.

  [0002] Digital video functions include digital television, personal digital assistant (PDA), laptop computer, desktop monitor, digital camera, digital recording device, digital media player, video game device, video game console, cellular telephone or satellite radio. It can be incorporated into a wide range of displays, including telephones, video teleconferencing devices, and the like. A display link is used to connect the display to the appropriate source device. Display link bandwidth requirements are proportional to the resolution of the display, and therefore high resolution displays require large bandwidth display links. Some display links do not have the bandwidth to support high resolution displays. Video compression can be used to reduce bandwidth requirements so that lower bandwidth display links can be used to provide digital video to high resolution displays.

  There are coding schemes that include image compression of pixel data. However, such schemes are sometimes not visually lossless or can be difficult and expensive to implement in conventional display devices.

  [0004] The Video Electronics Standards Association (VESA) has developed Display Stream Compression (DSC) as a standard for display link video compression. Display link video compression techniques, such as DSC, should especially provide picture quality that is visually lossless (ie, pictures with a level of quality that the user does not know compression is active). Display link video compression techniques should also provide a scheme that is easy and inexpensive to implement in real time using conventional hardware.

[0005] FIG. 1A is a block diagram illustrating an example video encoding and decoding system that may utilize techniques in accordance with aspects described in this disclosure. [0006] FIG. 1B is a block diagram illustrating another example video encoding and decoding system that may perform techniques in accordance with aspects described in this disclosure. [0007] FIG. 2A is a block diagram illustrating an example of a video encoder that may implement the techniques. [0008] FIG. 2B is a block diagram illustrating an example of a video decoder that may implement the techniques. [0009] FIG. 3 is a block diagram illustrating a search space for a non-first line for a 1D block. [0010] FIG. 4 is a block diagram illustrating a search space for a non-first line for a 2D block. FIG. 5 is a block diagram illustrating the search space for the first line for the 1D block. FIG. 6 is a block diagram illustrating the search space for the first line for 2D blocks. FIG. 7 is a flowchart illustrating a method for predicting a block of video data in a block prediction mode. [0014] FIG. 8 is a block diagram illustrating a block having partitions. [0015] FIG. 9 is a block diagram illustrating a data flow for a block prediction mode with an adaptive partition size. [0016] FIG. 10 is a block diagram illustrating two different partitioning options for a 2x2 region within a block. [0017] FIG. 11 is a block diagram illustrating an entropy coding group for a block prediction mode. FIG. 12 is a block diagram illustrating a search space for 2 × 8 blocks. [0019] FIG. 13 is a block diagram illustrating different partition sizes being used for different regions of the block. FIG. 14 is a flowchart illustrating a method for predicting a block of video data in a block prediction mode using a variable partition size. [0021] FIG. 15 is a block diagram illustrating an exemplary block prediction search for a 2x2 segment of 4: 2: 0 chroma subsampling. FIG. 16 is a block diagram illustrating an exemplary block prediction search for a 1 × 2 segment of 4: 2: 0 chroma subsampling. [0023] FIG. 17 is a block diagram illustrating an exemplary block prediction search for a 2x2 segment of 4: 2: 2 chroma subsampling. [0024] FIG. 18 is a block diagram illustrating an exemplary block prediction search for a 1x2 segment of 4: 2: 2 chroma subsampling. FIG. 19 is a block diagram illustrating a single search range for the block prediction mode. FIG. 20 is a block diagram illustrating a plurality of search ranges for the block prediction mode. FIG. 21 is a flowchart illustrating a method for predicting a block of video data in a block prediction mode using a plurality of search ranges. [0028] FIG. 22 is a block diagram illustrating an exemplary search region for a simplified block prediction mode. [0029] FIG. 23 is a block diagram illustrating an exemplary search region for a simplified block prediction mode. [0030] FIG. 24 is a block diagram illustrating an exemplary search region for a simplified block prediction mode. [0031] FIG. 25 is a block diagram illustrating an exemplary search region for a simplified block prediction mode. [0032] FIG. 26 is a flowchart illustrating a method for predicting a block of video data in a simplified block prediction mode.

  [0033] The DSC standard includes several coding modes in which each block of video data may be encoded by an encoder and similarly decoded by a decoder. In some implementations, the encoder and / or decoder may predict a current block to be coded based on a previously coded block.

  [0034] However, existing coding modes (eg, transform coding, differential pulse code modulation, etc.) do not provide a satisfactory way to compress highly complex regions in video data. Often, for this type of data (ie, highly compressed video data), the current blocks (or constituent sub-blocks) of the current block are coded by a coder (eg, encoder or decoder). The content is similar to the previous blocks encountered. However, existing intra prediction makes a satisfactory prediction of such a current block (eg, a prediction of the current block that is sufficiently similar to the current block and therefore will yield a sufficiently small residual). May be too limited. Accordingly, an improved method for coding a block of video data is desired.

  [0035] Each of the systems, methods and devices of the present disclosure has several inventive aspects, one of which is not solely responsible for the desired attributes disclosed herein. .

  [0036] In an aspect, a method for coding a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme may be used to predict a current block in a current slice. And the candidate block is within a range of pixel locations each corresponding to a reconstructed pixel in the current slice, and the range of pixel locations is at least (i) a plurality of pixel locations in the current slice A first region including one or more first pixel locations in a first line of pixels, wherein the first line of pixels includes at least one pixel in the current block (Ii) one or more second in the second line of the plurality of pixels in the current slice, spanning the entire width of the current slice A second region comprising pixel locations, wherein a second line of pixels comprises no pixels in the current block, but spans the entire width of the current slice, and comprises a plurality of pixel locations Determining a prediction vector that indicates a pixel position of a candidate block within a range of, and the pixel position of the candidate block is in one of the first region or the second region and signaling the prediction vector Coding the current block in a simplified block prediction mode, at least in part.

  [0037] In another aspect, an apparatus configured to code a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme includes one or more replays of a current slice of video data. A memory configured to store the configured pixels and one or more processors in communication with the memory may be included. One or more processors determine candidate blocks used to predict the current block in the current slice, and the candidate block is a plurality of pixels each corresponding to a reconstructed pixel in the current slice. A range of locations, wherein the range of pixel locations includes at least (i) a first region that includes one or more first pixel locations in a first line of the plurality of pixels in the current slice; Where the first line of pixels includes at least one pixel in the current block and spans the entire width of the current slice; (ii) in the second line of pixels in the current slice A second region including one or more second pixel locations, and wherein the second line of pixels does not include any pixel in the current block; Determining a prediction vector indicative of the pixel position of the candidate block within the range of the plurality of pixel positions, the pixel position of the candidate block being in the first region or second And coding the current block in a simplified block prediction mode, at least in part through signaling a prediction vector.

  [0038] In another aspect, the non-transitory physical computer storage may comprise code configured to code a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme. The code, when executed, determines to the device which candidate block is used to predict the current block in the current slice, and each candidate block corresponds to a reconstructed pixel in the current slice. A plurality of pixel positions within a plurality of pixel positions, the first including at least (i) one or more first pixel positions in a first line of the plurality of pixels in the current slice; And where the first line of pixels includes at least one pixel in the current block and spans the entire width of the current slice, (ii) the second of the pixels in the current slice A second region that includes one or more second pixel locations in the line, and wherein the second line of pixels includes any pixel in the current block. Determining a prediction vector indicative of the pixel position of the candidate block within the range of the plurality of pixel positions, the pixel position of the candidate block being in the first region or Coding the current block in a simplified block prediction mode, at least partly in one of the two regions and signaling a prediction vector.

  [0039] In another aspect, a video coding device may be configured to code a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme. The video coding device includes means for determining a candidate block used to predict a current block in the current slice, and the candidate block includes a plurality of pixel locations each corresponding to a reconstructed pixel in the current slice And the range of pixel locations is at least (i) a first region that includes one or more first pixel locations in a first line of pixels in the current slice; Where the first line of pixels includes at least one pixel in the current block and spans the entire width of the current slice; (ii) 1 in the second line of pixels in the current slice A second region including one or more second pixel locations, wherein a second line of pixels is any pixel in the current block The means for determining a prediction vector indicative of the pixel position of the candidate block within the range of the plurality of pixel positions, the pixel position of the candidate block comprising: Means for coding a current block in a simplified block prediction mode, at least partly in signaling in one of the region or the second region.

Detailed description

  [0040] In general, this disclosure relates to methods for improving video compression techniques, such as those utilized in display link video compression, for example. More particularly, this disclosure relates to systems and methods for coding blocks of video data in block prediction mode using adaptive search range selection.

  [0041] Although certain embodiments are described herein in the context of the DSC standard, which is an example of a display link video compression technique, the systems and methods disclosed herein may be any suitable video. Those skilled in the art will appreciate that it may be applicable to coding standards. For example, the embodiments disclosed herein include the following standards: International Telecommunication Union (ITU) Telecommunication Standardization Sector (ITU-T) H.264. 261, International Organization for Standardization / International Electrotechnical Commission (ISO / IEC) Moving Picture Expert Group 1 (MPEG-1) Visual, ITU-T H.264. 262 or ISO / IEC MPEG-2 Visual, ITU-T H.264. 263, ISO / IEC MPEG-4 Visual, ITU-T H.264 (also known as ISO / IEC MPEG-4 AVC). H.264, one or more of High Efficiency Video Coding (HEVC), and any extension to such a standard may be applicable. Also, the techniques described in this disclosure may become part of a standard that will be developed in the future. In other words, the techniques described in this disclosure may be applicable to previously developed video coding standards, currently developed video coding standards, and to the next video coding standard.

  [0042] The DSC standard includes several coding modes in which each block of video data can be encoded by an encoder and similarly decoded by a decoder. In some implementations, the encoder and / or decoder may predict a current block to be coded based on a previously coded block.

  [0043] However, existing coding modes (eg, transform coding, differential pulse code modulation, etc.) do not provide a satisfactory way to compress very complex regions in video data. Often, for this type of data (ie, highly compressed video data), the current block being coded (or a constituent sub-block of the current block) is the previous block encountered by the coder (eg, encoder or decoder). The content is similar. However, existing intra predictions are limited to making satisfactory predictions of such current blocks (eg predictions of current blocks that are sufficiently similar to the current block and therefore will yield sufficiently small residuals). It may have been done too much. Accordingly, an improved method for coding a block of video data is desired.

  [0044] In this disclosure, an improved method of coding a block in block prediction mode is described. For example, when searching for candidate blocks (or candidate regions) used to predict the current block (or the current region within the current block), the search range is good while the encoder minimizes the search cost Can be defined to have access to potential candidates that may be close matches. In another example, the encoder may determine which one of the multiple search ranges to use to code the current block based on a rate distortion (RD) analysis. In yet another example, the encoder is used to code the current block based on various factors such as the current block location, RD cost, etc., any one of the previously coded pixels. It can be determined whether it falls within the search range. Perform more operations at the encoder side (eg, search for candidate blocks used to predict the current block that may consume computing resources and processing power, candidates for the current block By calculating a vector that identifies the location of the block, comparing the associated cost with using different search ranges, etc., the method may reduce decoder complexity. Additionally, by allowing multiple and / or applicable search ranges to be used to code a block in block prediction mode, it may be more likely to be located in a better candidate partition. Thereby improving the coding efficiency and / or coding performance of the block prediction mode. Further, the performance of the block prediction scheme may be further improved by allowing an encoder for adaptively selecting a search range to be used to code each block.

Video coding standard
[0045] A digital image, such as a video image, a TV image, a still image, or an image generated by a video recorder or computer, may include pixels or samples arranged in horizontal and vertical lines. The number of pixels in a single image is typically tens of thousands. Each pixel generally includes luminance information and chrominance information. Without compression, the sheer amount of information to be conveyed from the image encoder to the image decoder would make real-time image transmission impractical. In order to reduce the amount of information transmitted, JPEG, MPEG and H.264. Several different compression methods have been developed, such as the H.263 standard.

  [0046] The video coding standard is ITU-T H.264. 261, ISO / IEC MPEG-1 Visual, and ITU-T H.264. 262 or ISO / IEC MPEG-2 Visual and ITU-T H.264. H.263, ISO / IEC MPEG-4 Visual, ITU-T H.264 (also known as ISO / IEC MPEG-4 AVC). H.264 and HEVC including extensions of such standards.

  [0047] Furthermore, a video coding standard, DSC, was developed by VESA. The DSC standard is a video compression standard that can compress video for transmission over a display link. As the display resolution increases, the bandwidth of the video data required to drive the display increases correspondingly. Some display links may not have the bandwidth to send all of the video data to the display for such resolutions. Thus, the DSC standard defines a compression standard for visual lossless compression that is interoperable over display links.

  [0048] The DSC standard is H.264. Different from other video coding standards such as H.264 and HEVC. DSC includes intra-frame compression but not inter-frame compression, which may not use temporal information by the DSC standard when coding video data. Means there is. In contrast, other video coding standards may employ interframe compression in their video coding techniques.

Video coding system
[0049] Various aspects of the novel systems, apparatus, and methods are described more fully hereinafter with reference to the accompanying drawings. However, this disclosure may be implemented in many different forms and should not be construed as limited to any particular structure or function presented throughout this disclosure. Rather, these aspects are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. Based on the teachings herein, the scope of the present disclosure, whether implemented in combination with any other aspect of the present disclosure, or in combination with any other aspect of the present disclosure, It should be understood by those skilled in the art that it is intended to cover any aspect of the novel systems, devices, and methods disclosed herein. For example, an apparatus may be implemented or a method may be performed using any number of aspects described herein. Further, the scope of the present disclosure may be implemented using other structures, functions, or structures and functions in addition to or in addition to the various aspects of the present disclosure described herein. It is intended to cover various devices or methods. It should be understood that any aspect disclosed herein may be embodied by one or more elements of a claim.

  [0050] Although particular aspects are described herein, many variations and permutations of these aspects fall within the scope of the disclosure. Although some benefits and advantages of the appropriate aspects are described, the scope of the disclosure is not intended to be limited to particular benefits, uses, or objectives. Rather, the aspects of the present disclosure are intended to be broadly applicable to different wireless technologies, system configurations, networks, and transmission protocols, some of which are illustrated by way of example in the figures and preferred aspects. This is illustrated in the following description. The detailed description and drawings are merely illustrative of the disclosure rather than limiting, the scope of the disclosure being defined by the appended claims and equivalents thereof.

  [0051] The accompanying drawings illustrate examples. Elements indicated by reference numerals in the accompanying drawings correspond to elements indicated by like reference numerals in the following description. In this disclosure, elements having names that begin with ordinal words (eg, “first”, “second”, “third”, etc.) do not necessarily imply that they have a particular order. Is not limited. Rather, such ordinal words are only used to refer to different elements of the same or similar type.

  [0052] FIG. 1A is a block diagram illustrating an example video coding system 10 that may utilize techniques in accordance with aspects described in this disclosure. The term “video coder” or “coder” as used and described herein generically refers to both video encoders and video decoders. In this disclosure, the terms “video coding” or “coding” may refer generically to video encoding and video decoding. In addition to video encoders and video decoders, aspects described in the present application include a transcoder (eg, a device that can decode a bitstream and re-encode another bitstream) and a middle box ( middlebox) (eg, a device that can modify, transform, and / or otherwise manipulate the bitstream) and can be extended to other related devices.

  [0053] As shown in FIG. 1A, video coding system 10 generates encoded video data that is subsequently decoded by destination device 14 (ie, “video coding device 14” or “coding device 14”). Source device 12 (ie, “video coding device 12” or “coding device 12”). In the example of FIG. 1A, the source device 12 and the destination device 14 constitute separate devices. However, it should be noted that source device 12 and destination device 14 may be on or part of the same device, as shown in the example of FIG. 1B.

  [0054] Referring once again to FIG. 1A, source device 12 and destination device 14 are each a telephone handset, such as a desktop computer, notebook (eg, laptop) computer, tablet computer, set-top box, so-called "smart" phone, or the like. Any of a wide range of devices (also called video coding devices), including so-called “smart” pads, televisions, cameras, display devices, digital media players, video game consoles, video streaming devices, and the like. In various embodiments, source device 12 and destination device 14 may be equipped for wireless communication (ie, configured to communicate via wireless communication).

  [0055] The video coding devices 12, 14 of the video coding system 10 may be via wireless networks and radio technologies such as a wireless wide area network (WWAN) (eg, cellular) and / or a wireless local area network (WLAN) carrier. Can be configured to communicate. The terms “network” and “system” are often used interchangeably. Each of video coding devices 12, 14 may be a user equipment (UE), a wireless device, a terminal, a mobile station, a subscriber station, and so on.

  [0056] WWAN carriers include, for example, code division multiple access (CDMA), time division multiple access (TDMA), frequency division multiple access (FDMA), orthogonal FDMA (OFDMA), single carrier FDMA (SC-FDMA), and others. Wireless communication networks, such as A CDMA network may implement a radio technology such as Universal Terrestrial Radio Access (UTRA), cdma2000, and so on. UTRA includes Wideband CDMA (WCDMA®) and other variants of CDMA. CDMA2000 covers IS-2000, IS-95 and IS-856 standards. A TDMA network may implement a radio technology such as Global System for Mobile Communications (GSM). OFDMA networks include Evolved UTRA (E-UTRA), Ultra Mobile Broadband (UMB), IEEE 802. 11 (Wi-Fi), IEEE 802. 16 (WiMAX), IEEE 802. 20, may implement a radio technology such as Flash-OFDM. UTRA and E-UTRA are part of the Universal Mobile Telecommunications System (UMTS). 3GPP® Long Term Evolution (LTE®) and LTE Advanced (LTE-A) are the latest releases of UMTS that use E-UTRA. UTRA, E-UTRA, UMTS, LTE, LTE-A and GSM are described in documents from an organization named “3rd Generation Partnership Project” (3GPP). CDMA2000 and UMB are described in documents from an organization named “3rd Generation Partnership Project 2” (3GPP2).

  [0057] Video coding devices 12, 14 of video coding system 10 may also be, for example, 802.11a-1999 (usually referred to as "802.11a"), 802.11b-1999 (usually "802.11b"). Communicate with each other over WLAN base stations according to one or more standards, such as the IEEE 802.11 standard, including modifications such as 802.11g-2003 (usually referred to as “802.11g”) Can do.

  [0058] Destination device 14 may receive encoded video data to be decoded via link 16. Link 16 may comprise any type of media or device capable of moving encoded video data from source device 12 to destination device 14. In the example of FIG. 1A, link 16 may comprise a communication medium to allow source device 12 to transmit encoded video data to destination device 14 in real time. The encoded video data may be modulated according to a communication standard such as a wireless communication protocol and transmitted to the destination device 14. The communication medium may comprise any wireless or wired communication medium, such as a radio frequency (RF) spectrum or one or more physical transmission lines. The communication medium may form part of a packet-based network, such as a local area network, a wide area network, or a global network such as the Internet. Communication media may include routers, switches, base stations, or any other equipment that may be useful for facilitating communication from source device 12 to destination device 14.

  In the example of FIG. 1A, source device 12 includes a video source 18, a video encoder 20 (also referred to simply as encoder 20), and an output interface 22. In some cases, output interface 22 may include a modulator / demodulator (modem) and / or transmitter. At source device 12, video source 18 may be a video capture device, such as a video camera, a video archive containing previously captured video, a video feed interface for receiving video from a video content provider, and / or source video. As a computer graphic system for generating computer graphic data, or a combination of such sources. As an example, if video source 18 is a video camera, source device 12 and destination device 14 may form a so-called “camera phone” or “video phone” as illustrated in the example of FIG. 1B. However, the techniques described in this disclosure may be generally applicable to video coding and may be applied to wireless and / or wired applications.

  [0060] Captured video, previously captured video, or computer-generated video may be encoded by video encoder 20. The encoded video data may be transmitted to the destination device 14 via the output interface 22 of the source device 12. The encoded video data may also (or alternatively) be stored on storage device 31 for later access by destination device 14 or other devices for decoding and / or playback. The video encoder 20 illustrated in FIGS. 1A and 1B may comprise the video encoder 20 illustrated in FIG. 2A, or other video encoder described herein.

  In the example of FIG. 1A, destination device 14 includes an input interface 28, a video decoder 30 (also referred to simply as decoder 30), and a display device 32. In some cases, input interface 28 may include a receiver and / or a modem. The input interface 28 of the destination device 14 may receive encoded video data via the link 16 and / or from the storage device 31. Encoded video data communicated over link 16 or provided on storage device 31 is generated by video encoder 20 for use by a video decoder, such as video decoder 30, in decoding the video data. Various syntax elements may be included. Such syntax elements may be included in encoded video data transmitted on a communication medium, stored on a storage medium, or stored on a file server. The video decoder 30 illustrated in FIGS. 1A and 1B may comprise the video decoder 30 illustrated in FIG. 2B, or other video decoder as described herein.

  [0062] The display device 32 may be integrated with or external to the destination device 14. In some examples, destination device 14 includes an integrated display device and may be configured to interface with an external display device. In other examples, destination device 14 may be a display device. In general, the display device 32 displays decoded video data to the user and can be used for various display devices such as a liquid crystal display (LCD), a plasma display, an organic light emitting diode (OLED) display, or another type of display device. Either can be provided.

  [0063] In a related aspect, FIG. 1B shows an example video coding system 10 ', where the source device 12 and the destination device 14 are on or part of the device 11. Device 11 may be a telephone handset, such as a “smart” phone. Device 11 may include a processor / controller device 13 that is in operative communication (optionally) with source device 12 and destination device 14. The video coding system 10 'of FIG. 1B and its components are otherwise similar to the video coding system 10 of FIG. 1A and its components.

  [0064] Video encoder 20 and video decoder 30 may operate according to a video compression standard, such as DSC. Alternatively, the video encoder 20 and the video decoder 30 are ITU-T H.264, alternatively called MPEG-4, Part 10, AVC. It may operate according to other proprietary or industry standards, such as H.264 standard, HEVC, or extensions of such standards. However, the techniques of this disclosure are not limited to any particular coding standard. Other examples of video compression standards are MPEG-2 and ITU-T H.264. H.263.

  [0065] Although not shown in the examples of FIGS. 1A and 1B, the video encoder 20 and video decoder 30 may each be integrated with an audio encoder and decoder, with audio in a common data stream or separate data streams. Appropriate MUX-DEMUX units, or other hardware and software, may be included to handle both video encodings. Where applicable, in some examples, the MUX-DEMUX unit is an ITU H.264 standard. It may be compliant with other protocols such as H.223 multiplexer protocol or User Datagram Protocol (UDP).

  [0066] Each of video encoder 20 and video decoder 30 includes one or more microprocessors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), discrete logic, software, It can be implemented as any of a variety of suitable encoder circuits, such as hardware, firmware, or any combination thereof. When the technique is implemented in part in software, the device stores instructions for the software in a suitable non-transitory computer readable medium and includes one or more processors to perform the techniques of this disclosure. Can be used to execute the instructions in hardware. Each of video encoder 20 and video decoder 30 may be included in one or more encoders or decoders, either of which may be integrated as part of a combined encoder / decoder at the respective device.

Video coding process
[0067] As briefly mentioned above, video encoder 20 encodes video data. Video data may comprise one or more pictures. Each of the pictures is a still image that forms part of the video. In some cases, a picture may be referred to as a video “frame”. When video encoder 20 encodes video data (eg, video coding layer (VCL) data and / or non-VCL data), video encoder 20 may generate a bitstream. A bitstream may include a sequence of bits that form a coded representation of video data. A bitstream may include coded pictures and associated data. A coded picture is a coded representation of a picture. The VCL data includes coded picture data (ie, information associated with a sample of coded picture (s)), and non-VCL data is a control associated with one or more coded pictures. Information (eg, parameter sets and / or supplementary extended information) may be included.

  [0068] To generate a bitstream, video encoder 20 may perform an encoding operation on each picture in the video data. When video encoder 20 performs an encoding operation on a picture, video encoder 20 may generate a series of coded pictures and associated data. The associated data may include a set of coding parameters such as a quantization parameter (QP). To generate a coded picture, video encoder 20 may partition the picture into equally sized video blocks. A video block can be a two-dimensional array of samples. Coding parameters may define coding options (eg, coding mode) for every block of video data. Coding options can be selected to achieve the desired RD performance.

  [0069] In some examples, video encoder 20 may partition a picture into multiple slices. Each of the slices may include spatially distinct regions in the image (eg, frame) that can be independently decoded without information from the rest of the regions in the image or frame. Each image or video frame can be encoded in a single slice, or each image or video frame can be encoded in several slices. In DSC, the number of bits allocated to encode each slice may be substantially constant. As part of performing the encoding operation on the picture, video encoder 20 may perform the encoding operation on each slice of the picture. When video encoder 20 performs an encoding operation on a slice, video encoder 20 may generate encoded data associated with the slice. Coded data associated with a slice may be referred to as a “coded slice”.

DSC video encoder
[0070] FIG. 2A is a block diagram illustrating an example of a video encoder 20 that may implement techniques in accordance with aspects described in this disclosure. Video encoder 20 may be configured to perform some or all of the techniques of this disclosure. In some examples, the techniques described in this disclosure may be shared between various components of video encoder 20. In some examples, additionally or alternatively, a processor (not shown) may be configured to perform some or all of the techniques described in this disclosure.

  [0071] For purposes of explanation, this disclosure describes video encoder 20 in the context of DSC coding. However, the techniques of this disclosure may be applicable to other coding standards or methods.

  [0072] In the example of FIG. 2A, video encoder 20 includes a plurality of functional components. The functional components of the video encoder 20 are a color-space converter 105, a buffer 110, a flatness detector 115, a rate controller 120, and a predictor. A quantizer and reconstructor component 125, a line buffer 130, an indexed color history 135, an entropy encoder 140, and a sub A stream multiplexor 145 and a rate buffer 150 are included. In other examples, video encoder 20 may include more, fewer, or different functional components.

  [0073] The color space converter 105 may convert the input color space to a color space used in a coding implementation. For example, in one exemplary embodiment, the input video data color space is in a red, green, and blue (RGB) color space and the coding is luminance Y, chrominance green Cg, and chrominance orange Co (YCgCo). Implemented in color space. Color space conversion may be performed by a method (s) that includes shifting and adding to the video data. Note that input video data in other color-spaces can be processed and conversion to other color spaces can also be performed.

  [0074] In a related aspect, video encoder 20 may include a buffer 110, a line buffer 130, and / or a rate buffer 150. For example, the buffer 110 may hold (eg, store) the color space converted video data prior to its use by other parts of the video encoder 20. In another example, color space converted data may require more bits, so video data may be stored in the RGB color space and color space conversion may be performed as needed.

  [0075] The rate buffer 150 may function as part of a rate control mechanism in the video encoder 20, which is described in more detail below with respect to the rate controller 120. The number of bits spent coding each block can vary greatly based substantially on the nature of the block. The rate buffer 150 can smooth out rate variations in the compressed video. In some embodiments, a CBR buffer model is employed in which bits stored in a rate buffer (eg, rate buffer 150) are removed from the rate buffer at a constant bit rate (CBR). In the CBR buffer model, the rate buffer 150 can overflow if the video encoder 20 adds too many bits to the bitstream. On the other hand, video encoder 20 may need to add enough bits to prevent underflow of rate buffer 150.

  [0076] On the video decoder side, bits may be added to the rate buffer 155 of the video decoder 30 (see FIG. 2B, described in further detail below) at a fixed bit rate, and the video decoder 30 may allow for each block. Variable bits may be deleted. In order to ensure proper decoding, the rate buffer 155 of the video decoder 30 should not “underflow” or “overflow” during decoding of the compressed bitstream.

[0077] In some embodiments, the buffer fullness (BF) is a value BufferCurrentSize that represents the number of bits currently in the buffer and the size of the rate buffer 150, ie, the rate buffer 150 at any point in time. It can be defined based on BufferMaxSize which represents the maximum number of bits that can be stored. BF can be calculated as follows.

  [0078] The flatness detector 115 can detect a change from a complex (ie, non-flat) area in the video data to a flat (ie, simple or uniform) area in the video data. The terms “complex” and “flat” are generally used herein to refer to the difficulty of video encoder 20 encoding each region of video data. Thus, the term complex as used herein generally indicates that the region of video data is complex for video encoder 20 to encode, eg, textured video data. , High spatial frequencies, and / or other features that are complex to encode. As used herein, the term flat generally indicates that the region of video data is simple for video encoder 20 to encode, eg, smooth slope in video data, low spatial frequency, and Other features that are simple to encode may be included. Transitions between complex and flat regions can be used by video encoder 20 to reduce quantization artifacts in the encoded video data. In particular, the rate controller 120, and the predictor, quantizer, and reconstructor components 125 may reduce such quantization artifacts when complex to flat region transitions are identified. Can do.

  [0079] The rate controller 120 determines a set of coding parameters, eg, QP. The QP is based on the buffer fullness of the rate buffer 150 and the image activity of the video data to maximize picture quality for the target bit rate that ensures that the rate buffer 150 does not overflow or underflow. Can be adjusted by. Rate controller 120 also selects a particular coding option (eg, a particular mode) for each block of video data to achieve optimal RD performance. The rate controller 120 minimizes the distortion of the reconstructed image so that the rate controller 120 satisfies the bit rate constraint, ie, the overall actual coding rate is within the target bit rate. Keep it down.

  [0080] The predictor, quantizer, and reconstructor component 125 may perform at least three encoding operations of the video encoder 20. The predictor, quantizer, and reconstructor component 125 may perform predictions in a number of different modes. One exemplary predication mode is a modified version of median-adaptive prediction. Median adaptive prediction may be implemented by the lossless JPEG standard (JPEG-LS). A modified version of the median adaptive prediction that may be performed by the predictor, quantizer, and reconstructor components 125 may allow parallel prediction of three consecutive sample values. Another exemplary prediction mode is block prediction. In block prediction, samples are predicted from previously reconstructed pixels that are on the top line or to the left of the same line. In some embodiments, video encoder 20 and video decoder 30 may both perform the same search on the reconstructed pixels to determine block prediction usage, so the bits are in block prediction mode. There is no need to be sent in. In other embodiments, video encoder 20 may perform a search and signal a block prediction vector in the bitstream so that video decoder 30 does not need to perform a separate search. A midpoint prediction mode in which samples are predicted using the midpoint of the component range may also be implemented. The midpoint prediction mode may allow the bounding of the number of bits required for compressed video even in the worst case sample. As discussed further below with reference to FIGS. 3-26, the predictor, quantizer, and reconstructor component 125 is based on video data based on one or more techniques described herein. May be configured to code (eg, encode or decode) a block (or any other unit of prediction). For example, the predictor, quantizer, and reconstructor component 125 may be configured to perform the method illustrated in FIG. 3-26. In other embodiments, the predictor, quantizer, and reconstructor component 125 may use one or more of the methods or methods described herein with one or more other components of video encoder 20. It may be configured to perform the technique.

  [0081] The predictor, quantizer, and reconstructor component 125 also performs quantization. For example, the quantization may be performed via a power-of-2 quantizer that may be implemented using a shifter. Note that other quantization techniques may be implemented instead of a power-of-two quantizer. The quantization performed by the predictor, quantizer, and reconstructor component 125 may be based on the QP determined by the rate controller 120. Finally, the predictor, quantizer, and reconstructor component 125 also adds an inverse quantized residual to the predicted value and ensures that the result is not outside the valid range of sample values. A reconfiguration including

  [0082] The exemplary techniques described above for prediction, quantization, and reconstruction performed by the predictor, quantizer, and reconstructor components 125 are merely examples, and other techniques are implemented. Note that it can be done. Note also that the predictor, quantizer, and reconstructor component 125 may include subcomponent (s) for performing prediction, quantization, and / or reconstruction. Further, it should be noted that prediction, quantization, and / or reconstruction may be performed by several separate encoder components instead of the predictor, quantizer, and reconstructor component 125.

  [0083] Line buffer 130 is a predictor, quantizer, and reconstructor component 125 and indexed color history 135 so that the predictor, quantizer and reconstructor component 125 and indexed color history 135 can use the buffered video data. The output from the quantizer and reconstructor component 125 is retained (eg, stored). The index color history 135 stores recently used pixel values. These recently used pixel values can be referenced directly by the video encoder 20 via a dedicated syntax.

  [0084] Entropy encoder 140 receives from predictor, quantizer, and reconstructor component 125 based on index color history 135 and flatness transitions identified by flatness detector 115. Encoded prediction residuals and any other data (eg, indices identified by the predictor, quantizer, and reconstructor component 125). In some examples, entropy encoder 140 may encode three samples per clock for each substream encoder. The substream multiplexer 145 may multiplex the bitstream based on a headerless packet multiplexing scheme. This allows video decoder 30 to operate three entropy decoders in parallel, facilitating decoding of three pixels per clock. Substream multiplexer 145 may optimize the packet order so that the packets can be efficiently decoded by video decoder 30. Note that different approaches to entropy coding may be implemented that may facilitate decoding of power-of-two pixels per clock (eg, 2 pixels / clock or 4 pixels / clock).

DSC video decoder
[0085] FIG. 2B is a block diagram illustrating an example of a video decoder 30 that may implement techniques in accordance with aspects described in this disclosure. Video decoder 30 may be configured to perform some or all of the techniques of this disclosure. In some examples, the techniques described in this disclosure may be shared between various components of video decoder 30. In some examples, additionally or alternatively, a processor (not shown) may be configured to perform some or all of the techniques described in this disclosure.

  [0086] For purposes of explanation, this disclosure describes video decoder 30 in the context of DSC coding. However, the techniques of this disclosure may be applicable to other coding standards or methods.

  [0087] In the example of FIG. 2B, the video decoder 30 includes a plurality of functional components. The functional components of video decoder 30 are rate buffer 155, substream demultiplexer 160, entropy decoder 165, rate controller 170, predictor, quantizer and reconstructor component 175, and index color history 180. A line buffer 185 and a color space converter 190. The illustrated components of video decoder 30 are similar to the corresponding components described above with respect to video encoder 20 in FIG. 2A. Accordingly, each of the components of video decoder 30 may operate in a manner similar to the corresponding component of video encoder 20 described above. In some embodiments, one or more components of video encoder 20 and / or video decoder 30 are one configured to execute software code configured to perform the tasks of such components. Or it may be implemented with multiple hardware processors. In other embodiments, one or more components of video encoder 20 and / or video decoder 30 may be implemented with hardware circuitry configured to perform the tasks of such components.

Slice in DSC
[0088] As described above, a slice generally refers to a spatially distinct region in an image or frame that can be independently decoded without using information from the rest of the region in the image or frame. Each image or video frame can be encoded in a single slice, or it can be encoded in several slices. In DSC, the target bits allocated to encode each slice may be substantially constant.

Block prediction mode
[0089] A single block of video data may include several pixels, and each block of video data has several potential coding modes in which the block may be coded. One such coding mode is the block prediction mode. In block prediction mode, the coder is in the previous reconstructed line (eg, if the current block is not in the first line of the current slice) or (eg, the current block is in the first line of the current slice). Attempts to find a candidate block in a previous reconstructed block in the same line that is close (eg, pixel value) to the current block being coded. In some embodiments, the proximity between pixel values is determined by a Sum of Absolute Differences (SAD) metric. The coder will find a candidate block in any part of the previously reconstructed block defined by the search range (which may be a predetermined value known to both encoder and decoder, for example). You can try. The search range is defined such that the encoder has potential candidates within the search range in order to find a good match while minimizing the search cost. The coding efficiency of the block prediction mode is such that if a good candidate is found (ie, a candidate in the search range whose pixel value is determined to be close to the current block being coded), it is between the candidate block and the current block. Comes from the fact that the difference (known as the residual) is smaller. A small residual requires fewer bits to signal compared to the number of bits required to signal the actual pixel value of the current block, thereby resulting in a lower RD cost, The possibility of being selected by the RD mechanism increases. For certain types of graphic content, the performance boost from enabling the block prediction mode is extremely significant.

Parameters in block prediction mode
[0090] The block prediction mode is designed to generate candidate blocks that provide the minimum distortion from the current block to be encoded given a specified search range. In some embodiments, the minimum distortion is defined using SAD. In some implementations of the present disclosure, the block prediction method is based on three parameters: a search range (SR), a skew (α), and a partition size (β). Defined. These three parameters affect the performance of the block prediction mode and can be adjusted (ie modified or reconfigured) during implementation. These parameters can be known to both the encoder and the decoder.

Search space in block prediction mode
[0091] In some embodiments of the present disclosure, the search space (eg, the spatial location of pixels that an encoder may search to find a candidate block) may vary based on the characteristics of the current block. While the search space may include all previously reconstructed blocks / pixels, the encoder and / or decoder may, for example, perform a search for candidate blocks to reduce computational complexity. It may be limited to a specified portion within the search space (eg, a “search scope” defined by one or more parameters that are either predefined or signaled in the bitstream). Examples of the block prediction search space are shown in FIGS. 3 and 4 illustrate the case with a current block that is not in the first line of the current slice (eg, current blocks 308 and 408). FIGS. 5 and 6 illustrate the case with a current block (eg, current blocks 506 and 606) that is in the first line of the current slice. These two cases are processed separately because the first line in the slice does not have a vertical neighbor. Thus, the reconstructed pixels from the current line can be utilized as search ranges (eg, search ranges 508 and 608). In this disclosure, the first line in the current slice may be referred to as FLS and any other line in the current slice may be referred to as NFLS.

  [0092] Further, the block prediction techniques described herein may be used for codecs that use a single line buffer (ie, 1D block size) or codecs that use multiple line buffers (ie, 2D block size). It can be implemented in either. Examples of search spaces for the 1D case are shown in FIGS. 3 and 5, and examples of search spaces for the 2D case are shown in FIGS. 4 and 6. For 2D, the search range is the pixel from the previous reconstructed line (eg, previous line 402) or the reconstructed block (eg, of current block 606) from the same line as the line in the 2D block. It may include the previous 604) in the current line 602, just to the left. A 2D block can be partitioned either horizontally or vertically or both. When accompanied by block partitions, a block prediction vector may be specified for each block partition.

Example implementation of block prediction mode
[0093] In some embodiments of the present disclosure, distortion metrics other than SAD, such as sum of squared difference (SSD), may be used. Alternatively or additionally, distortion can be corrected by weighting. For example, if the YCoCg color space is used, the cost can be calculated as follows:

  [0094] The block prediction techniques described herein may be performed in either the RGB color space or the YCoCg color space. In addition, the alternative implementation uses both color spaces, and one bit indicating which of the two color spaces is selected (eg, which color space has the lowest cost with respect to rate and distortion) The flag may be signaled to the decoder.

  [0095] In some embodiments of the present disclosure relating to FLS, one or more previous reconstructed blocks or blocks are excluded from the search range due to pipeline constraints and timing constraints. obtain. For example, depending on the hardware implementation, the coder may not complete the processing of the previous reconstructed block by the time the current block is processed by the coder (eg, replay for the previous block). The constructed pixel may not be known when the coder starts processing the current block), resulting in delay or failure. In such an implementation, by restricting the use of the previous reconstructed block to blocks where the reconstructed pixel value is known (eg, one or more previous reconstructed blocks). By excluding) the pipelining problem shown above can be solved. In some embodiments of the present disclosure for NFLS, the search range to the left of the current block may be from the same line, not the previous reconstructed line. In some of such embodiments, one or more previous reconstructed blocks may be excluded from the search range due to pipelining constraints and timing constraints.

Example implementation of NFLS
[0096] As shown in FIG. 3, the block prediction method may search the search range 310 (SR) in the search space to find candidates for the current block 308 (and the search of FIG. 4). The same applies to the space 400). If the x coordinate position of the first pixel of the current block 308 to be encoded is j, a set k of the starting positions of all candidate blocks in the search space may be given as follows:

  [0097] In this example, the parameter α skews the x coordinate position of the search range 310 relative to the current block to be encoded. Higher values of α shift search range 310 to the right, while lower values of α shift search range 310 to the left. For example, (i) 32 SRs and 15 α may place the search range 310 in the center of the previous line 302, and (ii) 32 SRs and 0 α will search the range 310 of the previous line 302. (Iii) 32 SRs and 31 α can place the search range 310 on the right side of the previous line 302.

  [0098] In some implementations of the present disclosure, a pixel that is within the search range but outside the slice boundary may be set to 1/2 of the dynamic range for that pixel. For example, if the content is RGB888, 128 default values for R, G, and B may be used. If the content is in YCoCg space, a default value of 128 may be used for Y and a default value of 0 may be used for Co and Cg (eg, Co and Cg are 9 centered on 0) Bit value).

Example implementation of FLS
[0099] As shown in FIG. 5, the search range may be different for the FLS case. This is because vertical neighbors are not available because such vertical neighbors are currently outside the frame, or because such vertical neighbors are included in different slices. In some embodiments of the present disclosure for the FLS case, the pixels in the current line may be used for block prediction. In one embodiment, any pixel in the current line to the left of the current block can be considered as part of the search range. In another embodiment, one or more previously coded blocks (eg, the previous block 504 immediately to the left of the current block) may be excluded from the search range due to pipelining constraints and timing constraints.

  [0100] In some implementations of FLS, the available range for the first few blocks in the first line of a slice will generally be less than the expected search range for other blocks. obtain. This is because the valid position for the candidate block starts at the beginning of the line and ends before the current block. For the first few blocks in the FLS, this effective range may be smaller than the desired range (eg, 32 or 64 positions). Thus, for these blocks, the search range may need to be adjusted so that each block segment of the candidate block is completely contained within the search range. For NFLS, the search range may be shifted to the left or right so that the total number of search locations is equal to the defined search range (eg, 32 or 64 pixel locations). Since j is the first pixel in the current block, the last pixel in the current block is j + blkWidth-1. For this reason, the search range may need to be shifted (blkWidth-1) pixels to the left.

  [0101] In some implementations of FLS, if the x coordinate location of the first pixel of the current block to be encoded is called j, the set of starting positions of all candidate blocks within the search range is Can be given.

(I) If the most recent previous reconstructed block is part of the search range, eg α = 1,

[Ii] (ii) If the n previous previous reconstructed blocks are excluded from the search range,

  [0104] Here, blkx is a block width. As described above for the NFLS case, any pixel outside the slice boundary can be set to a default value. Note also that there is no need to be associated when the skew parameter is FLS.

Exemplary flowchart for coding in block prediction mode
[0105] With reference to FIG. 7, an exemplary procedure for coding a block of video data in a block prediction mode is described. The steps shown in FIG. 7 may be performed by a video encoder (eg, video encoder 20 in FIG. 2A), a video decoder (eg, video decoder 30 in FIG. 2B), or component (s) thereof. Can be executed. For convenience, the method 700 is described as being performed by a video coder (also referred to simply as a coder), which may be the video encoder 20, the video decoder 30, or another component.

  [0106] The method 700 begins at block 701. In block 705, the coder determines candidate blocks that are used to predict the current block in the current slice. A candidate block may be within a range of locations (or pixel locations) defined by one or more block prediction parameters. For example, the block prediction parameters include (i) a search range parameter that defines the size of a range of locations, and (ii) a skew parameter that defines a relative location of the range of locations for the current block; (Iii) a partition size parameter defining the size of each partition in the current block. In some embodiments of the present disclosure, each of the search range parameter, the skew parameter, and the partition size parameter define a plurality of locations of candidate blocks in space rather than in time.

  [0107] At block 710, the coder determines a prediction vector based on the candidate block and the current block. The prediction vector may identify the location of the candidate block for the current block. The prediction vector may include one or more coordinate values (eg, coordinate values that indicate an offset in 1D space). At block 715, the coder codes the current block in block prediction mode, at least in part through signaling a prediction vector. In some embodiments, the coder may also signal the residual between the candidate block and the current block. Rather than having to signal the actual pixel value of the current block, a bit is signaled by signaling a prediction vector identifying the location of the candidate block and a residual representing the difference between the current block and the candidate block. Bit saving can be achieved. The method 700 ends at block 720.

  [0108] In method 700, one or more of the blocks shown in FIG. 7 may be deleted (eg, not performed) and / or the order in which the methods are performed may be reversed. . In some embodiments, additional blocks may be added to the method 700. The embodiments of the present disclosure are not limited to or by the example shown in FIG. 7, and other variations may be implemented without departing from the spirit of the present disclosure.

After finding candidate blocks
[0109] After the best candidate block is determined, the pixel value of the candidate block is subtracted from the pixel value of the current block, resulting in a residual. The residual may be quantized based on a preselected QP associated with the block prediction mode. The quantized residual is encoded using a codebook (which can be either fixed-length or variable-length) and can be either a fixed-length code or a variable-length code. May be signaled using. The selected codebook may be based on coding efficiency and hardware complexity requirements. For example, the selected codebook may be an Exp-Golomb codebook. In some embodiments of the present disclosure, an entropy coding scheme similar to delta size unit variable length coding (DSU-VLC) of existing DSC implementations may be used. In some embodiments, the residual may be transformed (eg, using a direct cosine transform, Hadamard transform, or other known transforms) prior to the quantization described above.

  [0110] In some embodiments of the present disclosure, the samples in the current block's residual may be partitioned into multiple groups (eg, 4 samples per group for a block containing 16 samples). . If all the coefficients in the block are 0, the block residual is coded using skip mode, i.e. to indicate whether the current component in the block is coded using skip mode. The 1-bit flag for each block (for each component) is signaled. Each group can be coded using DSU-VLC only if the group has one non-zero value if at least one non-zero value is included in the block. If a group (eg, 4 of the 16 samples in the residual) does not contain a non-zero value, the group is coded using skip mode, ie, the group uses skip mode. A 1-bit flag for each group is signaled to indicate whether it is coded. More particularly, for each group, a search can be performed to determine if all values in the group are zero. If all the values in the group are 0, a value of “1” may be signaled to the decoder; otherwise (if at least one value is not 0), a value of “0” is signaled to the decoder; Followed by coding of DSU-VLC coding. In the alternative, a value of “0” may be signaled if all values in the group are zero, and a value of “1” may be signaled if the group contains at least one non-zero value.

  [0111] In some embodiments of the present disclosure, the best candidate block is explicitly signaled to the decoder by transmitting a fixed length code that includes the best offset. The offset may be referred to as a “vector”. The advantage of explicitly signaling the vector to the decoder is that the decoder does not need to perform the block search itself. Rather, the decoder explicitly receives the vector and adds the candidate block to the decoded, dequantized residual value to determine the pixel value of the current block.

Block classification
[0112] In some embodiments of the present disclosure, the current block to be coded is partitioned, resulting in multiple candidate blocks and multiple vectors per block. In some of such embodiments, the vector (s) may be explicitly signaled using a fixed length code. For example, the length of this fixed length code can be log ₂ (SR). In another embodiment, the vector (s) may be explicitly signaled using a variable length code, such as a code from an exponential Golomb or Golomb-Rice code family. This codebook may be selected based on a statistical distribution associated with the vector (s). In yet another embodiment, the vector (s) can be predicted based on the previously coded vector (s), and the residual of the vector (s) is It can be coded using any fixed or variable length code. In yet another embodiment, the vector (s) can be predicted based on the previously coded vector (s) to signal whether the two vectors are the same. 1-bit flags can be used. This flag can be called SameFlag. If SameFlag = 1, the vector value itself need not be signaled to the decoder. If SameFlag = 0, the vector is explicitly signaled (eg, using either a fixed length code or a variable length code). An exemplary block partitioning scheme is shown in FIG.

  [0113] As shown in diagram 800 of FIG. 8, current block 802 includes a single partition. The information signaled for the current block 802 comprises a mode header, vector SameFlag, vector A, and payload. The current block 804 includes two sections, section A and section B. The information signaled for the current block 804 comprises a mode header, vector SameFlag, vector A, vector SameFlag, vector B, and payload. As explained above, one or more of the items listed above may not be signaled. For example, if the vector SameFlag is equal to 1, subsequent vectors need not be signaled.

  [0114] The partition size β may determine the partition of the current block into separate sub-blocks. In such cases, a separate block prediction may be performed for each sub-block. For example, if the block size is N = 16 and the partition size β = 8β = 8, the search is performed for each of 16/8 = 2 partitions. In another example, if β = N, the block partition is disabled. If β <N, each vector can be explicitly signaled to the decoder. If vector prediction is not employed (eg, using previously signaled vectors to define the current vector), each vector is signaled using a fixed or variable length code. If vector prediction is employed, the first vector may be predicted (eg, stored in memory) from the previous coded vector, and for n> 0, vector n is predicted from vector n-1.

Variable partition size in block prediction mode
[0115] The above examples have a size of 1x8 (eg, having a height of 1 pixel and a width of 8 pixels), or 2x8 (eg, having a height of 2 pixels and a width of 8 pixels). Fig. 6 illustrates how a block (with) can be coded in block prediction mode. As shown in FIG. 8, a block can be partitioned into multiple regions, each region using a different partitioning scheme (eg, using a 1 × 2 partition and using a 2 × 2 partition). And so on) and a block prediction vector may be specified for each partition (eg, signaled in the bitstream with the residual associated with each partition). For example, each block may be partitioned into multiple 1 × 2 partitions that include two pixels (or other fixed size partitions).

  [0116] In other embodiments, the encoder may determine the most efficient block partition size for each block (for each sub-region within the block). Efficiency can be measured based on the rate and distortion associated with coding a block (or sub-regions therein) using a given block partition size. For example, when coding a block that includes four 2 × 2 regions, the encoder uses a single partition (eg, a single 2 × 2 partition for each 2 × 2 region) using the first three Determine that maximum coding efficiency can be achieved by coding a 2 × 2 region and coding a fourth 2 × 2 region using two partitions (eg, two 1 × 2 partitions) Can do. By allowing the encoder to select the partition size for each block as applicable, the performance of the block prediction scheme may be further improved. This can be used to smooth large regions (eg, regions that show no change in pixel values across the region or less than a change in threshold amount), so that the block prediction vector This is because it requires fewer bits to signal (e.g., related to the size of the region), while using a smaller segmentation (reducing distortion and / or entropy coding rate reduces additional signaling costs). Can be used for complex areas (weighting). For example, the encoder may determine whether a given region or block satisfies a smoothing threshold condition, and in response to determining that a given region or block satisfies a smoothing threshold condition, The size may be used to encode a given region or block in block prediction mode (in other cases, a smaller partition size may be used to encode a given region or block in block prediction mode). As another example, an encoder may determine whether a given region or block satisfies a complexity threshold condition, and in response to determining that a given region or block satisfies a complexity threshold condition, Encode a given region or block in block prediction mode using a smaller partition size (otherwise encode a given region or block in block prediction mode using a larger partition size) Can do. The ability to select different partition sizes adaptively may allow block prediction modes to be used with a wider range of content types (eg, graphic content, natural images, test patterns, fine text rendering). .

Exemplary data flow for coding in block prediction mode
[0117] FIG. 9 illustrates an example data flow 900 for coding a block in block prediction mode using an adaptive partition size. As illustrated in FIG. 9, the current block 902 predicted in the block prediction mode includes a block partition 904. In one example, the block partition has a size of 1 × 2 or 2 × 2. A block prediction (BP) search 906 is performed to identify blocks or partitions that are available and already coded to predict the current block 902 (or block partition 904) in block prediction mode. . As shown in FIG. 9, the BP search 906 codes the current line containing the current block, for example, the previous line (eg, immediately preceding line or another preceding line). One or more previous reconstructed blocks 907A in the line coded prior to doing) and / or previous reconstructed blocks from the current line (eg, the line containing the current block) The search range including 907B may be searched.

  [0118] The encoder determines the block predictor 908 based on the candidate block or partition identified in the search range. Block predictor 908 is subtracted from current block 902 (or current block partition 904 in candidate block 902) at block 910, and the residual determined based on the subtraction is quantized at block 912. The quantized residual is entropy coded by an entropy coder 920. In addition, inverse quantization 914 is performed on the quantized residual and the result is applied to block predictor 908 at block 916 to generate a reconstructed block 918. The BP partition size selection 922 is performed based on the distortion performance (D) of the reconstructed block 918 and the rate performance (R) of the entropy encoded residual. Bitstream 924 is generated based on the selected BP partition size.

  For example, the BP partition size selection 922 receives the rate (eg, R) and distortion (eg, D) of each partition region (eg, 2 × 2) in the current block 902 as input, and the partition region is simply Should be coded using a single block prediction vector (BPV) (eg, a total of 1 BPV for a single 2 × 2 partition) or multiple BPVs (eg, for two 1 × 2 partitions) Whether to be partitioned and coded using a total of 2 BPV, each of which is 1 BPV, may be determined for prediction based on the RD tradeoff between the two options. Some examples discussed herein include a 2 × 2 partition region size (thus having 1 × 2, 2 × 1, and 2 × 2 partition sizes as selectable options), but an encoder The partition sizes selectable by are not limited to those used in such examples (eg, 1 × 2 and 2 × 2), but other sizes (eg, 2 × based on block size and / or region size). X1).

  [0120] In some embodiments, the partition size is fixed in the current partition region or block (eg, 1 × 2, 2 × 2, or any other sub-combination of pixels. For example, a block is The block size may be 2 × 8, and the block may be divided into sub-blocks or regions having a size of 2 × 2, 2 × 2 sub-blocks or regions within the 2 × 8 block are 1 × 2 In such an example, each 1 × 2 partition can be predicted using a single BPV, independent of other partitions, in other embodiments, the partition The size is variable, and which partition size is used to code each block, sub-block, and / or region in block prediction depends on the rate and distortion of each partition scheme For example, for a 2 × 2 region (eg, the current region) in the current block, predicting the current region by dividing the current region into two 1 × 2 partitions, and Predicting two 1 × 2 partitions using two BPVs separately (eg, each directed to a previously coded 1 × 2 partition within a defined search range) ( The current region may be predicted using a 1 × 2 partitioning scheme if it yields better rate and / or distortion performance (as compared to other partitioning schemes such as 2 × 2, for example) (eg, Predicting the current region as a single 2 × 2 partition using one BPV (indicating a previously coded 2 × 2 partition) within the defined search range (eg, 1 × 2 Other division like The current region can be predicted using a 2 × 2 partitioning scheme if it yields better rate and / or distortion performance (compared to the equation) The partitioning scheme used to code the blocks in prediction mode The process of determining is described in more detail below with reference to FIG.

Block size and sub-block size
[0121] For an M × N block size, some embodiments provide subblocks of size M _sub × N _sub (also referred to herein as regions), where M _sub ≦ M and N _sub ≦ N. Reference is made to the description. In some implementations, both M _sub and N _sub are aligned to an entropy coding group in an M × N block to simplify computation. Each sub-block M _sub × N _sub within a block is predicted using a single BPV without further partitioning, using the BPV used for each partition, or ( ii) It can be either divided into multiple sections (eg, two 1 × 2 sections). An efficient trade-off between using a single BPV for the entire sub-block, or partitioning a sub-block into partitions each with its own BPV is a bit that signals more BPV However, using more BPV may reduce distortion and entropy coding rates, which can result in extra rates in the stream. In other words, using more bits to signal additional BPV reduces the number of bits used to signal the residual (difference between candidate block / region and current block / region) However, it also causes the number of bits used for entropy coding to decrease as well. The encoder compares each option (eg, no partition vs. multiple partitions) for RD cost and selects whether to partition each sub-block or region based on the cost comparison, or the best One partitioning scheme may be selected from a plurality of partitioning schemes that provide RD performance.

Example classification scheme
[0122] FIG. 10 illustrates a diagram 1000 illustrating an exemplary partitioning scheme. FIG. 10 illustrates two partitioning options for a 2 × 2 sub-block or region. In this example, block 1002 (eg, including pixels X ₀ -X ₁₅ ) has a size of 2 × 8, and sub-block or region 1004 (eg, pixels X ₀ , X ₁ , X in the block). ₈ and X ₉ ) has a size of 2 × 2. Partition option 1006 illustrates an example where a single BPV is used to predict a sub-block or region 1004, and partition option 1008 includes two BPVs for each 1 × 2 partition in the sub-block or region 1004. Illustrates an example where a sub-block or region 1004 is predicted using. In some implementations such as Advanced DSC (Adv-DSC) to align sub-blocks or regions to the entropy coding group structure 1100 for block prediction mode shown in FIG. A subblock or region is used. In the example of FIG. 11, entropy coding groups 0, 1, 2, and 3 are illustrated, each corresponding to one of the four 2 × 2 sub-blocks or regions in the block. However, the techniques described herein are not limited to such embodiments and can be extended to any block size M × N and any sub-block size M _sub × N _sub . However, in the example illustrated below, parameters M = 2, N = 8, M _sub = 2 and N _sub = 2 are used. In some embodiments, the sub-block and / or partition scheme may be determined based on an entropy coding group. For example, the sub-block and / or partition scheme may be determined such that each sub-block and / or partition scheme is included in a single entropy coding group.

Determine partition size
[0123] The encoder may either (i) code each 2x2 region as a single 2x2 partition, or (ii) divide the region into two 1x2 partitions, each based on the minimum RD cost It may be decided whether to code the 1 × 2 partition separately. The RD cost can be calculated as shown below.

[0124] In some implementations, the BPV is signaled with a fixed number of bits (BPV _bits ) equal to log ₂ (SR) log ₂ (SR), where SR is associated with the block prediction mode. Search space (or search range). For example, if the search space consists of 64 locations, log ₂ (64) = 6 bits are used to signal each BPV.

[0125] The search space for block prediction using variable partition sizes may be slightly different from the search range discussed with reference to FIGS. In particular, the M _sub × N _sub sub-block can use a search space using the height M _sub . In such a case, an additional line buffer may be required to implement block prediction using a variable partition size versus block prediction that does not use a variable partition size. An example of such a search space is demonstrated in FIG. 12 for a 2 × 2 sub-block size. FIG. 12 illustrates a diagram 1200 illustrating an exemplary search range. As shown in FIG. 12, current line 1202 includes (i) current block 1204 with current sub-block 1206 and (ii) previous block 1208. In the example of FIG. 12, the previous line 1210 includes a search range 1212 from which the encoder can select a candidate sub-block 1214 to predict the current sub-block 1206. The search range or space for a 1D partition (eg, 1 × 2) is similar to the search range described above with reference to FIG. 3 and may depend on a single previous reconstructed line .

[0126] In some embodiments, the distortion D _{2 × 2} and D _{1 × 2} can be calculated using a modified sum of absolute differences in YCoCg color space (SAD). For example, the SAD distortion between pixel A (eg, in the current subblock or partition) and pixel B (eg, in a candidate subblock or region) in the YCoCg color space may be calculated as follows:

  [0127] If the current sub-block or partition has more than one pixel, the distortion for the entire current sub-block or partition is obtained by summing the individual SADs calculated for each pixel in the current sub-block or partition. Can be calculated. The pixel value of the current sub-block or partition can be an actual pixel value or a reconstructed pixel value (eg, calculated based on a candidate predictor and a residual). In some implementations, the lambda parameter may be fixed at a value of two. In other implementations, this parameter may be tuned depending on block size, bit rate, or other coding parameters.

[0128] Entropy coding costs EC _bits may be calculated for each 2x2 region. The four samples in each entropy coding group are either 2 × 2 quantized residuals predicted from a single BPV (eg 2 × 2 partitions), or 2 vectors (eg 2 1 × 2 partitions) ) Can be derived from any of the 2 × 2 quantized residuals. For example, the entropy coding cost may represent the number of bits required to signal each entropy coding group in the bitstream (eg, including vector (s) and residual). Based on the calculated entropy coding cost, the encoder may select the partitioning scheme with the lowest cost for each 2 × 2 region. Some embodiments are discussed with reference to a 2 × 8 block having a 2 × 2 sub-block size, a 2 × 2 entropy coding group, and two partitioning schemes (1 × 2 and 2 × 2). The techniques described herein may be extended to other block sizes, sub-block sizes, entropy coding groups, and / or partition schemes.

Signaling coding information in the bitstream
[0129] In the 2 × 8 block 1002 shown in FIG. 10, each of the four 2 × 2 regions may be partitioned based on the RD cost analysis described above. For example, each 2 × 2 region may be partitioned into either a single 2 × 2 partition or two 1 × 2 partitions. Four examples of such partitions are illustrated in FIG. 1300 of FIG. As shown in FIG. 13, block 1302 has four sub-blocks predicted based on the 2 × 2 partition scheme, and block 1304 includes three predicted subblocks based on the 2 × 2 partition scheme. A sub-block and one sub-block predicted based on the 1 × 2 partition scheme, a block 1306 includes four sub-blocks predicted based on the 1 × 2 partition scheme, and a block 1308 includes It has one sub-block predicted based on the 2 × 2 partition and three sub-blocks predicted based on the 1 × 2 partition scheme. In addition to signaling the BPV to the decoder, the encoder may also send one bit for each 2x2 region so that the decoder can properly guess the partition. In some implementations, such as the Adv-DSC implementation, a group of 4 bits indicating the partitioning scheme selected for each region in the block (eg, each 2 × 2 region in a 2 × 8 block) is a bit Signaled in the stream. In such an implementation, the four bits “1011” indicate that the first, third, and fourth regions (eg, 2 × 2 sub-blocks) in the block are based on the first partitioning scheme (eg, 1 The second region (eg, 2 × 2 sub-block) is predicted or coded based on the second partition scheme (eg, based on 2 × 2 partition) or It can indicate that it is coded. In some embodiments, following these four bits in the bitstream, the BPV may be signaled using a fixed bit for each BPV. In the previous example (eg, a bit sequence of “1011”), 7 BPV may be signaled.

Exemplary flowchart for coding in block prediction mode
[0130] With reference to FIG. 14, an exemplary procedure for coding a block of video data in a block prediction mode is described. The steps illustrated in FIG. 14 may be performed by a video encoder (eg, video encoder 20 of FIG. 2A), or their component (s). For convenience, the method 1400 is described as being performed by a video coder (also referred to simply as a coder), which may be the video encoder 20, or another component.

  [0131] The method 1400 begins at block 1401. At block 1405, the coder uses one or more first schemes to predict a current region (eg, in a block of video data coded in block prediction mode). The candidate area is determined. For example, the first candidate region may be one of 2 × 2 regions in a 2 × 8 block. The first partitioning method may be a partitioning method in which the current region is partitioned into a plurality of partitions (for example, two 1 × 2 partitions). In some embodiments, the one or more first candidate regions are within a first range of locations associated with the first partitioning scheme (eg, a search range associated with the first partitioning scheme). It is in. The one or more first candidate regions may be stored in a memory of the video encoding device.

  [0132] At block 1410, the coder determines one or more second candidate regions that are used to predict the current region using the second partitioning scheme. For example, the second partition scheme may be a partition scheme in which the current region is not partitioned into multiple partitions (eg, the current region is coded as a single 2 × 2 partition). In another example, the second partitioning scheme may be a partitioning scheme in which the current region is partitioned into a number of partitions that is different from the number of partitions used for the first partitioning scheme. In some embodiments, the one or more second candidate regions are within a second range of locations associated with the second partition scheme (eg, a search range associated with the second partition scheme). It is in. One or more second candidate regions may be stored in a memory of the video encoding device.

  [0133] At block 1415, the coder may use a first cost associated with coding the current region using the first partitioning scheme to code the current region using the second partitioning scheme. Is determined to be greater than a second cost associated with. For example, the coder was associated with coding a current region using a rate and distortion based on the rate and distortion associated with coding the current region using a first partitioning scheme, and a second partitioning scheme. Costs based on rate and distortion can be calculated and the calculated costs can be compared.

  [0134] At block 1420, the coder receives the second one or more prediction vectors identifying the location of one or more second candidate regions for the current region, at least in part. Code the current region using a partitioning scheme. The method 1400 ends at block 1425.

  [0135] In the method 1400, one or more of the blocks shown in FIG. 14 may be deleted (eg, not performed) and / or the order in which the methods are performed may be reversed. In some embodiments, additional blocks may be added to the method 1400. The embodiments of the present disclosure are not limited to or by the example shown in FIG. 14, and other variations may be implemented without departing from the spirit of the present disclosure.

Extensions to 4: 2: 0 and 4: 2: 2 chroma subsampling formats
[0136] In some implementations, the block prediction techniques described in this disclosure (eg, using variable partition sizes in block prediction mode) may be utilized only for 4: 4: 4 chroma sampling formats. . This format is mainly used for graphic content. For example, the 4: 4: 4 chroma sampling format utilizes image or video data that includes color components (eg, luma and chroma components) having the same sampling rate (eg, not using chroma subsampling). However, the 4: 4: 4 chroma sampling format may be used less frequently for other video applications. Because of the significant compression that chroma subsampling can provide, both 4: 2: 0 and 4: 2: 2 chroma subsampling formats are primarily used for video applications. For example, some versions of DSC (eg, DSCv1.x) may support 4: 2: 0 and 4: 2: 2. Support for such chroma subsampling formats may be utilized or required in future DSC implementations. Thus, in some embodiments, the block prediction techniques described in this disclosure (eg, using a variable partition size in block prediction mode) are in 4: 2: 0 and / or 4: 2: 2 format. Expanded. Although 4: 2: 0 and 4: 2: 2 chroma subsampling formats are used herein, the various techniques described in this application may be applied to other known sampling formats.

  [0137] In some embodiments, the algorithm for block prediction using a variable partition size works greatly in the same way, independent of the chroma sampling format. In such an embodiment, regardless of the format (eg, 4: 4: 4, 4: 2: 2, 4: 2: 0, etc.), a single partition (eg, 2 × 2) is used, Or a determination of whether to use multiple partitions (eg, two separate 1 × 2 partitions), or the number of partitions used to code the current sub-block or region (eg, 1, 2, 3, 4) may be made for each sub-block or region (eg, 2 × 2 block) of luma samples. However, the number of chroma samples in each partition or in each block can vary depending on the subsampling format. In addition, the encoder decision is modified in 4: 2: 2 and / or 4: 2: 0 chroma subsampling format because alignment with entropy coding groups may no longer be available for chroma components. You may need that. Thus, for encoder decisions (eg, when the encoder decides whether to divide each 2 × 2 region into a single 2 × 2 partition or two 1 × 2 partitions based on the minimum RD cost) The rates for each partition (eg, rate values associated with a partition such as a single 2 × 2 partition or two separate 1 × 2 partitions) are for 4: 2: 2 and 4: 2: 0 Can depend only on luma samples. For example, when calculating SAD distortion, any terms associated with the chroma component (s) may be set to zero.

BP search for 4: 2: 0 chroma subsampling format
[0138] With respect to the 2x2 partition in 4: 2: 0 mode (4: 2: 0 chroma subsampling format), each partition is single for each of the chroma components (eg, Co and Cg, or Cb and Cr). One chroma sample may be included. In some embodiments, the chroma samples used (eg, to calculate RD costs and / or predict samples in the current region or block) are those that intersect the partition. . In other embodiments, the chroma samples used can be derived from adjacent segments. An exemplary 2 × 2 search 1500 for 4: 2: 0 mode is shown in FIG. In FIG. 15, chroma sites (eg, sample / pixel locations with chroma samples) are indicated using “X”. For example, the upper left sample of section A, the upper right sample of section B, and the upper left sample of the current section comprise chromasites that intersect each section. Such chromasites can be used for all calculations performed for each segment (eg, to calculate a difference value using chroma sample values).

  [0139] For the 1x2 segment in 4: 2: 0 mode, there is no chromasite in the second line of the current block, so the 1x2 segment in the first line of the current block and the current block A distinction may need to be made between the 1 × 2 sections in the second line. For example, for the segment in the first line of the current block, the distortion value calculation may include two luma samples and one chroma sample for each chroma component. For the segment in the second line of the current block, the distortion value calculation may include only luma samples (eg, two luma samples). In the example 1600 of FIG. 16, the current 1 × 2 segment A is in the first line and includes chromasite. Thus, the candidate segment selected to predict the current 1 × 2 segment A is the candidate 1 × 2 segment A, which also includes chromasite. Similarly, the current 1 × 2 segment B is in the second line and does not contain chromasite. Thus, the candidate segment selected to predict the current 1 × 2 segment B is the candidate 1 × 2 segment B, which also does not include chromasite.

BP search for 4: 2: 2 chroma subsampling format
[0140] For a 2x2 partition in 4: 2: 2 mode (4: 2: 2 chroma subsampling format), each partition has four luma samples and chroma components (eg, Co and Cg, or Cb and Two chroma samples for each of (Cr). An exemplary 2 × 2 search 1700 for 4: 2: 2 mode is shown in FIG. In FIG. 17, chroma sites (eg, pixel locations with chroma samples) are indicated using “X”. For example, the two left samples of section A, the two right samples of section B, and the two left samples of the current section comprise chromasites that intersect their respective sections. Such chromasites can be used for all calculations performed for each segment (eg, to calculate a difference value using chroma sample values).

  [0141] For a 1x2 segment in 4: 2: 2 mode, each segment includes two luma samples and one chroma sample for each of the chroma components (eg, Co and Cg, or Cb and Cr). including. Unlike the 4: 2: 0 mode, there may not be a distinction between the section in the first line of the current block and the section in the second line of the current block in the 4: 2: 2 mode. is there. An exemplary block prediction search 1800 for a 1 × 2 partition for 4: 2: 2 chroma subsampling is illustrated in FIG. In the example of FIG. 18, the current 1 × 2 segment A is in the first line, the current 1 × 2 segment B is in the second line, and each of the current segments A and B has a chromasite. Including. Current segment A is predicted based on candidate 1 × 2 segment A, which includes chromasite in the first sample, and current segment B is predicted based on candidate 1 × 2 segment B, which is In the second sample, chromasite is included. Thus, regardless of whether the chromasite is located within the candidate partition, the chroma sample can be used to predict the chroma sample in the current partition.

Encoder determination
[0142] In 4: 2: 2 and 4: 2: 0 formats, there may be less than 4 entropy coding groups per block for each chroma component. For example, four entropy coding groups may be used for luma components, two (or one) entropy coding groups may be used for orange chroma components, and two (or one). An entropy coding group may be used for the green chroma component. The number of entropy coding groups used to code a given block may be determined based on the number of luma or chroma samples in a given block. In some embodiments, the entropy coding group is determined by the encoder based on the coding mode in which a given block is coded. In other embodiments, the entropy coding group is configured with an applicable coding standard (eg, based on the coding mode in which a given block is coded).

[0143] In some embodiments, the quantity EC _bits is not necessarily accurately determined by the encoder for the chroma. In some of such embodiments, based on entropy coding rates calculated using only luma samples for 4: 2: 2 and 4: 2: 0 formats, the encoder is 1 × 2 It may be decided whether to use a partition or a 2 × 2 partition. In other embodiments, the quantity EC _bits is determined by the encoder for the chroma, and to the entropy coding rate calculated using both luma and chroma samples for 4: 2: 2 and 4: 2: 0 formats. Based on this, the encoder may decide whether to use a 1 × 2 partition or a 2 × 2 partition.

Signaling
[0144] In some embodiments, the number of entropy coding groups transmitted from the encoder to the decoder for each block or for each color component may be varied depending on the chroma subsampling format. In some implementations, the number of entropy coding groups is changed to ensure that the codec throughput is high enough. For example, in 4: 4: 4 mode, a 2 × 8 block may include four entropy coding groups as illustrated in FIG. In such an example, four entropy coding groups may be used (eg, signaled by an encoder) for each color component (eg, Y, Co, and Cg). Table 1 illustrates exemplary changes to the number of entropy coding groups used for 4: 2: 2 and 4: 2: 0 modes. The remainder of signaling described above (eg, BPV signaling, segmented indication signaling, etc.) is for 4: 2: 2 and 4: 2: 0 modes (for 4: 4: 4 mode). May not change) (from the signaling described). For example, in Table 1, component 0 may correspond to luma (Y), component 1 may correspond to orange chroma (Co), and component 2 may correspond to green chroma (Cg).

advantage
[0145] One or more block prediction mode techniques described in this disclosure may be implemented using an asymmetric design. The asymmetric design allows more expensive procedures to be performed on the encoder side and reduces decoder complexity. For example, since the vector (s) are explicitly signaled to the decoder, the encoder does most of the work compared to the decoder. This is desirable because the encoder is often part of a system-on-chip (SoC) design that operates at high frequencies on state-of-the-art process nodes (eg, 20 nm or less). On the other hand, the decoder can be implemented with a display driver integrated circuit (DDIC) chip-on-glass (COG) solution that has a limited clock speed and a much larger process size (eg, greater than 65 nm). There is sex.

  [0146] Furthermore, the adaptive selection of block partition size allows the block prediction mode to be used for a wider range of content types. Since explicit signaling of BPV is expensive, a variable partition size allows for reduced signaling costs for image regions that can be well predicted using 2 × 2 partitions. For highly complex regions, if the entropy coding rate can be reduced sufficiently to compensate for higher signaling costs, or the distortion is reduced sufficiently so that the RD trade-off further favors 1 × 2. If so, a 1 × 2 partition size may be selected. For example, an adaptive selection of block partition sizes can increase performance across all content types, including natural images, test patterns, fine text rendering, and the like. In some embodiments, the adaptive partitioning techniques described herein may be extended by considering a block partition size greater than 2x2 and / or a block size greater than 2x8.

  [0147] One or more techniques described herein may be implemented in a fixed-bit codec using a fixed bit rate buffer model. Bits stored in such a model, rate buffer, are deleted from the rate buffer at a fixed bit rate. Thus, if the video encoder adds too many bits to the bitstream, the rate buffer can overflow. On the other hand, the video encoder may need to add enough bits to prevent underflow of the rate buffer. Further, on the video decoder side, bits can be added to the rate buffer at a fixed bit rate, and the video decoder can delete a variable number of bits for each block. To ensure proper decoding, the video decoder rate buffer should not "underflow" or "overflow" during decoding of the compressed bitstream. One or more techniques described herein may ensure that no such underflow or overflow occurs during encoding and / or decoding. In some embodiments, the encoder may operate under a bit-budget constraint, where the encoder is a fixed number of bits to code a given region, slice, or frame. Have In such an embodiment, how many bits of each one of the multiple coding modes are given so that the encoder can ensure that other bits / bandwidth associated with the bit budget or constraints can be met. Being able to know exactly (without having to estimate) whether it is necessary to be able to code a given region, slice or frame is critical to the encoder. For example, if the coding of a given region, slice, or frame requires more bits than estimated, the encoder can perform the given coding without having to perform any preliminary measurements. A given region, slice, or frame may be coded in mode.

  [0148] Furthermore, one or more techniques described herein overcome certain technical problems associated with video compression techniques in transmission over a display link. A video encoder by allowing a region to be coded based on a plurality of candidate regions (eg, each partition in a region predicted based on a corresponding one of the plurality of candidate regions) The decoder can provide customized predictions based on the nature of the region (eg, smooth, complex, etc.), thereby improving video encoder and decoder (eg, hardware and software codec) performance.

Multiple search ranges for block prediction mode
[0149] As described with reference to FIGS. 3-6, the search space (eg, the spatial location of pixels that an encoder may search to find a candidate block) varies based on the characteristics of the current block. obtain. For example, the search space may potentially include all previously reconstructed blocks / pixels. In some embodiments, the encoder and / or decoder may perform a search for a candidate block, for example, in a specified portion in the search space (e.g., predefined in the bitstream, to reduce computational complexity, for example. Or “search range” defined by one or more parameters that are either signaled or signaled. In some implementations, block prediction utilizes a single search range for each block coded in block prediction mode. In these implementations, the location of the search range for the current block may depend on whether the current block is at FLS (first line of the slice) or NFLS (non-first line of the slice). As shown in diagram 1900 of FIG. 19, if the current block 1910 is FLS, the search range may be to the left of the current block in the same block line (eg, FLS search range 1920) and the current block is If in NFLS, the search range may be in the block line immediately above the current block line (eg, NFLS search range 1930). The term block line may include all raster scan lines belonging to a block in addition to its normal meaning. For example, if the block size is 2 × 8 pixels (Advanced Display Stream Compression (ADSC) has a standard block size of 2 × 8 pixels), the block line can be two raster scan lines. Would include.

  [0150] In contrast, in some embodiments of the present disclosure, an encoder and / or decoder may maintain multiple search ranges. By allowing multiple search ranges to be used to code a block in block prediction mode, it may be more likely to be located in a better candidate partition (eg, coded in block prediction mode). Compared to previous implementations that only consider a single search range for each block), thereby improving the coding efficiency and / or coding performance of the block prediction mode. Furthermore, the performance of the block prediction scheme can be further improved by allowing an encoder to select a search range to be used to code each block.

  [0151] In some of such embodiments, multiple search ranges may be considered for use in coding a given block in block prediction mode, but only one of the search ranges. May be able to be used at certain times. For example, each block coded in block prediction mode may be associated with one of the plurality of search ranges instead of both. In some embodiments, if a block coded in block prediction mode has multiple partitions, the coding of those partitions is coded using the same search range selected for the block. Can be constrained. By limiting the number of search ranges used for a single block, the encoder can easily signal to the decoder which search range is used using a single bit. . In other embodiments, more than one search range may be used for a single block. For example, a first search range may be used to code a first partition in a single block, and a second search range that is different from the first search range is a first search range in a single block. Can be used to code two partitions.

[0152] In some embodiments of the present disclosure, the two search ranges (SR ₀ and SR ₁ ) are maintained by an encoder and / or decoder, as shown in FIG. 2000 of FIG. Because the only option is to use the current block line for reference for blocks in the FLS, there may not be a distinction between the two search ranges (or differences in block prediction results or performance) May not exist). For example, if the current block 2010 is in the FLS, only the SR ₀ search range 2020 may be used to code the current block 2010, and the SR ₀ search range 2030 may not be available. On the other hand, if the current block 2010 is in the NFLS, both the SR ₀ search range 2020 and the SR ₁ search range 2030 may be available, and one of the search ranges 2020 and 2030 is in the block prediction mode. Can be used to code 2010. As illustrated in FIG. 20, the SR ₀ search range 2030 includes data (eg, of the current block) from a previous reconstructed block line (eg, the most recently reconstructed block line or lines). SR ₁ search range 2020 includes data from the current block line (eg, to the left of the current block) (eg, pixels that were coded prior to coding of the current block). Including. In some embodiments, one or more most recently reconstructed pixels or blocks may be removed from the search range (s) from pipelining reasons. For example, one or more blocks (eg, pixels or threshold number of blocks) that are immediately to the left of the current block may be removed from the search range SR ₁ . The number of pixels or blocks removed from the search range (s) may depend on pipelining constraints.

  [0153] The encoder may perform a block prediction search for all partitions in the current block, independently of the two search ranges. For example, if the current block has two partitions, the encoder may perform a block prediction search for the first partition in the first search range, and then for the second partition in the first search range. A block prediction search may be performed. Based on the search, the encoder may determine a first cost for coding two partitions in the current block using multiple blocks or multiple block partitions in a first search range. The encoder may then perform a block prediction search for the first partition in the second search range and then perform a block prediction search for the second partition in the second search range. Based on this search, the encoder may determine a second cost for coding the two partitions in the current block using multiple blocks or multiple block partitions in the second search range. A block or block partition in the search range may be selected so that rate and distortion costs are minimized for the entire current block (eg, to predict all partitions in the current block).

  [0154] Having determined the cost (eg, rate and distortion estimation) for each search range, the encoder has two options by minimizing the RD cost (eg, D + λ · R) as discussed in this disclosure. You can choose between. The encoder may select the search range that results in the lowest RD cost and use the selected search range to code the current block. The search range indication used to decode the current block is sent to the decoder, eg, by explicitly signaling a 1-bit flag for each block in the bitstream. Thus, the decoder side changes required by the use of multiple search ranges are minimal. In essence, one search range is replaced with the other, and all other steps for block prediction can be performed as in an implementation that does not use multiple search ranges.

  [0155] In other embodiments, the 1-bit flag signaling the selected search range may be omitted. In such an embodiment, the search range may each be associated with a separate instance of the block prediction mode, where the search range index may be implicitly signaled by the mode header. For example, if 3 bits are used to signal the coding mode associated with the block, only 6 coding modes are available to the encoder or decoder, and the same 3 bit syntax element is used for 2 additional coding modes. (For example, one always uses the first search range or the block prediction mode that uses the first search range by default, and the other always uses the second search range or defaults. In block prediction mode using the second search range). Thus, bit savings can be achieved by utilizing existing syntax elements used to signal the coding mode.

Coding in block prediction mode using multiple search ranges
[0156] With reference to FIG. 21, an exemplary procedure for coding a block of video data in a block prediction mode is described. The steps shown in FIG. 21 may be performed by a video encoder (eg, video encoder 20 in FIG. 2A), or component (s) thereof. For convenience, the method 2100 is described as being performed by a coder, which may be the video encoder 20, or another component.

  The method 2100 begins at block 2101. At block 2105, the coder associates with coding a current block (eg, a block of video data that is currently coded) based on a first candidate region within a first range of locations corresponding to the current block. Determine the first cost given. The first candidate region may have the same size (eg, the same dimensions and / or the same number of pixels) as the current block. The first candidate region may include a previously coded block or portion of a block that is currently used to code the current block. In some embodiments, the first candidate region may be a collection of blocks or block partitions that are each used to code different portions of the current block. For example, the current block may include four block partitions, each of the four block partitions being predicted or coded using a different block or block partition of the first candidate region within the first range of locations. obtain. In some implementations, multiple block partitions within the current block may be coded based on the same block or block partition of the first candidate region within the first range of multiple locations. The first range of multiple locations (eg, the first search range) may be a search range specified by an encoder or by an applicable coding standard. The first range of locations may be similar to one of the exemplary search ranges discussed in this disclosure. The first range of locations is a plurality of blocks that are reconstructed and used to predict or code subsequent blocks and / or block partitions (eg, in coding order or raster scanning order) Or a block section may be provided. The first range of locations may include a raster scan line that overlaps the current block. In other embodiments, the first range of locations does not include raster scan lines that overlap the current block. Video data associated with the first candidate region may be stored in a memory of the video encoding device.

  [0158] At block 2110, the coder determines a second cost associated with coding the current block based on a second candidate region within a second range of a plurality of locations corresponding to the current block. To do. The second candidate region may have the same size (eg, the same dimensions and / or the same number of pixels) as the current block. The second candidate region may include one previously coded block or part of a block and is currently used to code the current block. In some embodiments, the second candidate region may be a set of blocks or block partitions that are each used to code different portions of the current block. For example, the current block may include four block segments, each of the four block segments being predicted using different blocks or block segments of the first candidate region within the second range of the plurality of locations. Can be coded. In some implementations, multiple block partitions within the current block may be coded based on the same block or block partition of the first candidate region within the second range of multiple locations. The second range of locations may be a search range specified by an encoder or by an available coding standard. The second range of locations may be similar to the exemplary search range discussed in this disclosure. The second range of locations is a plurality of blocks that are reconstructed and used to predict or code subsequent blocks and / or block partitions (eg, in coding order or raster scanning order) Or a block section may be provided. In some embodiments, the first range of locations and the second range of locations are mutually exclusive. Alternatively or additionally, the first range of locations and the second range of locations may occupy different raster scan lines. The second range of locations may include raster scan lines that overlap the current block. In other embodiments, the second range of locations does not include raster scan lines that overlap the current block. For example, as shown in FIG. 20, the two search ranges 2020 and 2030 do not overlap each other. Video data associated with the second candidate region may be stored in a memory of the video encoding device.

  [0159] At block 2115, the coder associates a first cost associated with coding the current block based on the first candidate region to code the current block based on the second candidate region. Determine if it is greater than the given second cost. For example, a coder may cost based on rate and distortion associated with coding a current block using a first candidate region within a first range (eg, within a first search range) of multiple locations. And a rate and distortion-based cost associated with coding the current block using a second candidate region within a second range (eg, within a second search range) of the plurality of locations. These calculated costs can then be compared. In some embodiments, the current block may comprise multiple block partitions. In some of such embodiments, calculating the first and second costs includes (i) an associated search range used to code a corresponding plurality of block partitions in the current block. Determining a plurality of block segments within (eg, a first search range and a second search range, respectively), and (ii) individual blocks within the current block based on the plurality of block segments within the associated search range Determining individual costs to code a plurality of block sections and (iii) calculating first and second costs based on the individual costs. For example, the first and second costs can be calculated by summing the individual costs. Alternatively, the first and second costs can be calculated by averaging the individual costs.

  [0160] In block 2120, the coder at least partially provides providing an indication associated with the second range in response to determining that the first cost is greater than the second cost. And coding the current block based on the second candidate region within the second range of the plurality of locations. In some embodiments, the indication may be that the current block is coded (i) based on a first candidate region within a first range of multiple locations, or (ii) a second of multiple locations. Can be a 1-bit flag that indicates whether to code based on a second candidate region in the range. For example, if the flag value is equal to 0, the flag indicates that the current block is one or more in the first search range (eg, based on a first candidate region in the first range of locations). If the flag value is equal to 1, the flag may indicate that the current block is based on a second candidate region (eg, in a second range of locations). And) may indicate coding based on one or more blocks or block partitions in the second search range. In other embodiments, the indication may be a multi-bit syntax element configured to indicate a coding mode associated with the current block. For example, the syntax element may indicate which one of a plurality of coding modes should be used to code the current block. One of the coding modes may be a block prediction mode. In some embodiments, if the syntax element has one value (of multiple possible values), the current block uses only the first search range (or others). If not coded, the current block is coded in block prediction mode (which uses the first search range by default) and the syntax element has another value (among several possible values). Coded in a block prediction mode that uses only two search ranges (or uses the second search range by default unless otherwise provided). If the syntax element may have yet another value (among multiple possible values), the current block may be coded in a coding mode other than the block prediction mode. The method 2100 ends at block 2125.

  [0161] In the method 2100, one or more of the blocks shown in FIG. 21 may be deleted (eg, not executed) and / or the order in which the methods are executed may be interchanged. . For example, in some embodiments, one or more of blocks 2105, 2110, and 2115 may be any preceding raster scan line in the same slice (eg, the first line in the slice). May be omitted if the coder determines that the current block contains raster scan lines that do not have. In some embodiments, additional blocks may be added to method 2100. The embodiments of the present disclosure are not limited to or by the example shown in FIG. 21, and other variations may be implemented without departing from the spirit of the present disclosure.

Advantages of using multiple search ranges
[0162] Techniques related to using multiple search ranges when coding a block in block prediction mode improve the coding efficiency associated with block prediction mode, and in particular, graphic type images and graphics. Increase content coding performance. Implementing one or more of these techniques may increase computational complexity at the encoder side. However, since the encoder is implemented at a smaller process node (20 nm or less), the encoder typically exhibits a greater degree of tolerance to increased computational complexity. Importantly, the decoder complexity will remain largely the same even when multiple search ranges are used to code the block in block prediction mode. The decoder can generally be implemented with a much larger process size (60 nm or more) and can follow more stringent hardware requirements (eg, gate count must be minimized). Thus, the techniques of this disclosure for using multiple search ranges in block prediction mode improve coding performance with a relatively small increase in computational complexity.

Simplified block prediction mode
[0163] In some cases, the techniques described above for coding the current block in block prediction mode can be further simplified. For example, for a cost-constrained hardware implementation, one or more features described above reduce the computational complexity of the coder (on the encoder side, the decoder side, or both) Can be deleted or modified. In such cases, one or more of the following changes can be made to code the block in block prediction mode without significantly degrading performance: (i) The coder is described above. Instead of using multiple search ranges, a single search range may be used to predict the current block or partition; (ii) the search range may be a previous reconstructed line (eg, Includes pixels from both the current line and the current line, where the samples in such a line have already been reconstructed (eg, the time the current block or partition is coded) And / or (iii) a single previously reconstructed line is an alternative to using a previous reconstructed block line (which may include multiple lines) , It is used to predict the current block or segment.

  [0164] Depending on the desired trade-off between a given implementation between coding performance and hardware complexity, the block prediction modes described herein (eg, standard block prediction modes, multiple ranges) Various versions and modifications of the techniques for coding blocks in block prediction modes using, simplified block prediction modes, etc. may be used. Some versions of the block prediction mode may be selected for ADSC depending on the VESA task group promise between performance and hardware complexity.

[0165] As described above, in some embodiments, the simplified block prediction mode may use a single search range. In some of such embodiments, the total number of possible block prediction vectors is determined as 2 ⁿ for some n. For example, ADSC typically uses n = 6, in which case the total number of possible block prediction vectors would be 64 positions. Candidate pixels within the search range may come from any of three regions that may be referred to herein as region A, region B, and region C. An exemplary mapping of the BVP index to the search range (SR) and a position within the search range (SR pos) is shown in Table 2. For example, this mapping may be computed from the associated SR length SrLen _i , iε {A, B, C}.

[0166] In some embodiments, the block prediction vector that the encoder explicitly signals to the decoder may be an integer in the range [0, 2 ⁿ -1]. The mapping from the index to the search range may depend on SrLen _i . Table 2 shows an example where SrLen _A = 26, SrLen _B = 8, and SrLen _C = 30.

[0167] In Figure 2200 of Figure 22, a single search range comprising pixels from different regions of a causally-available image (eg, previously reconstructed pixels) is simplified. An example used by generalized block prediction is shown. The number of candidates in each particular region may be adjusted depending on the codec parameters. In the example of FIG. _22, SR A / SR _B is formed from the previous reconstructed line, whereas, SR _C is formed from the current block line. For example, SR _A is directly above the current block 2340, as illustrated in FIG. 22 (e.g., perpendicular thereto direction (a vertically) overlapping) comprises, or right of the pixel either, SR _B is , Including the pixel to the left of the current block 2340 as illustrated in FIG. 22 (eg, does not overlap vertically with the current block 2340 and has a smaller x coordinate value than the pixels in the current block 2340) . FIG. 22 illustrates SR _A 2220, SR _B 2210, SR _C 2230, and current block 2240. As illustrated in FIG. 22, SR _A 2220 and SR _B 2210 are in the previous reconstructed line, and SR _C 2230 is currently in the block line.

FIG. 2300 of FIG. 23 illustrates an example of a simplified block prediction mode using a variable partition size (2 × 2). Search in SR _C, as described herein (e.g., by determining the cost associated with coding the current block using the 2 × 2 block in SR _C) are performed obtain. For searches in SR _A / SR _B , the candidate partition may be expanded or padded in the y direction to create 2 × 2 candidates. FIG. 23 illustrates SR _A 2320, SR _B 2310, SR _C 2330, and current block 2340. In FIG. 23, SR _A 2320 and SR _B 2310 are in the previous reconstructed line, and SR _C 2330 is currently in the block line.

FIG. 2400 of FIG. 24 illustrates an example of a simplified block prediction mode using a variable partition size (1 × 2). Search in SR A _/ SR _B, as described herein (e.g., determining a cost associated with coding the current block using the 1 × 2 block in SR A _/ SR _B Can be performed). For SR C search _(current block line) in, classified in the current line l of the block is searched from the line l of SR _C. FIG. 24 illustrates SR _A 2420, SR _B 2410, SR _C 2430, and current block 2440. In FIG. 24, SR _A 2420 and SR _B 2410 are in the previous reconstructed line, and SR _C 2430 is currently in the block line.

[0170] For example, the number of search positions for a particular region (eg, region A, B, or C) may be referred to herein as SrLen _i for region i. In such an example, the following constraints may be established: SrLen _A + SrLen _B + SrLen _C ≦ 2 ⁿ . For example, if block prediction is performed using a single search range and the maximum number of positions in a single search range is defined to be 2 ⁿ , the sum of the positions in each of those regions is It will need to be below the maximum number. The value for SrLen _i may be adjusted depending on the needs of the codec. In addition, these values can be easily adjusted dynamically based on the partition in the current slice or the location of the current block. For example, if the current block or partition is located in the FLS, the encoder and decoder may infer that SR _A and SR _B cannot be used to code the current block. Therefore, the number of positions in the SR _C (e.g., up to the maximum value assigned to a single search range) may be allocated.

[0171] In addition to or as an alternative to using a single search range, the requirement for the encoder / decoder to store the previous reconstructed block line in a simplified block prediction mode is removed. obtain. Instead, only one previous reconstructed line can be stored. For example, for any block size P × Q, only one reconstructed line (included in the search range as SR _A in and 24) that may be stored in place of the P line.

  [0172] If variable piecewise sizing is leveraged, subsequent logic changes may be implemented for the simplified block prediction mode.

[0173] In some implementations, if a 2x2 partition is used to code the current block, any candidate position from SR _A / SR _B has been described above with reference to FIG. As such, it can be expanded or padded in the y direction to generate 2 × 2 candidates. For example, as illustrated in FIG. 23, a 1 × 2 candidate 2350 may be expanded or padded in the y direction by duplicating sample values to generate a 2 × 2 candidate 2360. Similar techniques can be extended to blocks of any size. For example, candidates can be expanded or padded in the y direction to match the height of the current block or partition. On the other hand, the 2 × 2 candidate 2380 may be used as not expanded or padded. In other implementations, how the 2 × 2 partition in the current block is coded depends on which search range (eg, SR _A , SR _B , or SR _C in FIGS. 22-24) It may depend on what is used for coding. Such a technique is described in more detail below with reference to FIG.

[0174] 1 if × 2 segment is used to code the current block, none of the candidate position from SR _C, as described above with reference to FIG. 24, the current of the current block It can be selected from the same line as the 1 × 2 section. For example, as illustrated in FIG. 24, current segment 2450 is predicted based on 1 × 2 candidates 2460 in the same line, and current segment 2470 is predicted based on 1 × 2 candidates 2480 in the same line. The In such an example, to find a candidate for current partition 2450, the coder codes current partition 2450 based on individual 1 × 2 blocks in search range 2430 in the same line as current partition 2450. To compare the costs and find candidates for the current partition 2470, the coder costs the coding of the current partition 2470 based on the individual 1 × 2 blocks in the search range 2430 in the same line as the current partition 2470. Compare

Benefits of coding in simplified block prediction mode
[0175] Techniques for coding in a simplified block prediction mode provide a trade-off between performance and complexity on both the encoder side and the decoder side. This is desirable for any implementation that is constrained in hardware costs.

Further simplification of simplified block prediction mode
[0176] Further modifications to the search range used in the simplified block prediction mode described above may be made to reduce the area of ADSC implementation for ASIC / FPGA. The hardware implementation of the ADSC decoder may require fast random access to all locations within the search range. For example, such a hardware implementation may include an array of flip-flops that are proportional to the size of the search range (eg, in the worst case). Thus, it may be desirable to limit the maximum number of possible locations within each part of the search range (eg, the size of the region within the search range). In one example, the maximum number of possible locations within each region within the search range may be as follows: SR _A = 20, SR _B = 12, SR _C = 32. For example, the number of positions in each region may be limited to such maximum number regardless of how many positions are in other regions. If the current block being processed is at location x = 128 in the first line of the slice (eg, having 128 pixels before the current block in the same line), search ranges A and B code the current block Additional pixels can be included in the search range C without having available pixels to do so and not exceeding the maximum size of the search range (e.g., if the maximum size of the search range is 64 pixels, Despite the fact that 64 of the 128 previously coded pixels may be included in the search range, the number of positions for search range C may be limited to 32. Such limits can be placed at the expense of coding efficiency to limit the amount of storage required in hardware. From the encoder perspective, the other 32 search range positions (eg, the first 20 pixels and the last 12 pixels of the 64 position search ranges) are relative to any current block in the first line of the slice. It can be “invalid”. In some implementations, each part of the search range is always assigned to the same number of positions, and each position is available to the encoder whether there is a pixel at that position or at the time of coding the current block. Depending on whether there is, it can be “valid” or “invalid”. Block predictive search and all other operations (eg, cost calculation and comparison) may be skipped for such invalid locations. The number of valid positions increases to the right edge of the first line of the slice (eg, as illustrated by the second column in FIG. 25, where the search range 2520 is extended to subsequent block lines). Will do. In other implementations, the sum of the number of positions at each position in the search range may be limited to a maximum number (eg, 64 positions) or less. In such an implementation, if the current block being processed is at position x = 128 in the first line of the slice (eg, having 128 pixels before the current block in the same line), the search range The number of positions for C can be equal to greater than 32 (eg, up to 64 if the maximum number is 64), since other search ranges (eg, A and B) can be empty.

  [0177] To limit the impact on coding efficiency, under certain circumstances, a smaller number of bits may be used to signal a block prediction vector in the bitstream. For example, if the current block is within a certain position (eg, the first line of a slice), both the encoder and decoder are used to signal a block prediction vector with a smaller number of bits. And using block prediction vectors that are signaled using less than the number of bits required to accurately identify each individual position within the search range (e.g., search range 6 bits), the candidate block or partition can be accurately identified. In the above example where 32 of the 64 positions are determined to be “invalid”, only 32 of the 64 search range positions are valid during that time, so instead of 6, block prediction Five bits per vector may be used in the majority of the first line of the slice.

  [0178] In addition, with respect to block timing, the ability to fill the search range flip-flops at a constant rate may be advantageous for hardware implementation of ADSC. This means that the search range should be efficiently shifted by one block width every block time. As a result, certain positions within the search range C may technically be in the previous block line with respect to the current block once the current block has advanced to the next line of the slice. An illustration of this feature is shown in FIG. When the current block 2510 moves to the right edge of the slice and then moves to the next block line, the search range 2530 (eg, the portion of the search range within the current block line) is the fourth and It remains in the previous block line, as shown in the 5 rows.

  [0179] In some embodiments, the search range B (eg, the top line of the search range 2540 of FIG. 25) generates 2 × 2 prediction candidates that span the line before the current block line and the first line. Can be used to As shown in FIG. 25, the top line of search range 2540 is search range B, and the bottom line of search range 2540 is a portion of search range C that is collocated with respect to search range B (eg, search range 2530). Thus, in some of such embodiments, instead of extending or padding the 1 × 2 prediction candidates in search range B to generate 2 × 2 prediction candidates, the coder may Two pixels from the captured line (eg, two pixels from the search range B) and two pixels from the first line of the current block line (eg, two pixels in the search range B) , Two pixels from the search range C). This approach is not available when the pixels from the current block line that are collocated with respect to the pixels in search range A (eg, directly below the pixels in search range A) cause the current block 2510 coding time. Therefore, it cannot be used for search range A (eg, search range 2520).

Coding in block prediction mode with simplified search range
[0180] With reference to FIG. 26, an exemplary procedure for coding a block of video data in a block prediction mode will be described. The steps shown in FIG. 26 may be performed by a video encoder (eg, video encoder 20 in FIG. 2A), or component (s) thereof. For convenience, the method 2600 is described as being performed by a coder, which can be the video encoder 20, or another component.

  [0181] The method 2600 begins at block 2601. At block 2605, the coder determines candidate blocks that are used to predict the current block in the current slice, where the candidate block is a plurality of pixels each corresponding to a reconstructed pixel in the current slice. It is within the range of the position (for example, search range). For example, the coder determines a cost associated with coding the current block based on each potential candidate block of the plurality of potential candidate blocks in a range of pixel locations and has the lowest cost Is identified as a candidate block. Each potential candidate block may correspond to one of the pixel locations within a plurality of pixel locations. The range of pixel locations may include a first region that includes one or more first pixel locations in a first line of pixels in the current slice, where the first of the plurality of pixels The line overlaps the current block. For example, the first line of pixels may span the entire width of the current slice, and the first line of pixels may include at least one pixel in the current block. Further, the plurality of pixel location ranges includes a second region that includes one or more second pixel locations in a second line of pixels in the current slice, wherein a second of the plurality of pixels. This line does not overlap with the current block. For example, the second line of pixels can span the entire width of the current slice and does not include any pixels in the current block. The second line of pixels may immediately precede the first line in the current slice. Each of the first and second lines may be a raster scan line within the current slice. In some embodiments, the first region and the second region occupy different raster scan lines. The first region may be in a raster scan line that overlaps the current block (eg, the raster scan line and the current block include at least one common pixel). The range of pixel locations may further include a third region that includes one or more third pixel locations in the second line (eg, in the same line that includes the second region). For example, one or more third pixel locations in the third region are collocated with respect to pixel locations in the first line that are part of the current block (or overlap vertically with the current block). May not include any pixel location in the second line, while at least one of the one or more second pixel locations in the second region is one of the current block. It may include one or more second pixels in the second line that are collocated with respect to pixel locations in the first line that are parts (or that overlap vertically with the current block). As described herein, a region may each include a different number of pixel locations. For example, the number of pixel locations in the first region is greater than the number of pixel locations in the second region, which has a greater number of pixel locations than the third region. In some embodiments, the current block is a 1 × 2 partition within a 2 × 8 block predicted in a simplified block prediction mode. In other embodiments, the current block is a 2 × 2 partition within a 2 × 8 block predicted in a simplified block prediction mode. In some other embodiments, the current block is a 2 × 8 block predicted in a simplified block prediction mode. Each potential candidate block in the range of pixel locations may be any pixel location in the range of pixel locations (eg, a plurality of pixels in the first region, second region, or third region). Position) (eg, included as an upper left pixel or another reference pixel). Video data associated with the candidate block may be stored in a memory of the video encoding device.

  [0182] At block 2610, the coder determines a prediction vector that indicates the pixel location of the candidate block within the range of pixel locations. For example, the pixel location of the candidate block may be in one of the first region or the second region.

  [0183] At block 2615, the coder codes the current block in a simplified block prediction mode, at least in part through signaling a prediction vector. The coder may signal the prediction vector using a fixed number of bits (eg, the minimum number of bits required to uniquely identify each pixel location in the range of pixel locations). For example, if there are 64 pixel locations within a plurality of pixel locations, 6 bits can be used to signal each prediction vector. In some embodiments, if the location of the current block in the current slice prevents the range of pixel positions from having more than a certain number of pixel positions that is less than the maximum number of pixel positions, the coder May signal the prediction vector using less than the number of bits required to uniquely identify the maximum number of pixel locations within the range of pixel locations. For example, the coder may not have a range of pixel locations having more than 32 pixel positions due to the location of the current block in the current slice (eg, the current line is the first in the current slice). If there are only 32 reconstructed blocks that precede the current block in raster scan order), the reduced number of bits (eg, 5 in this case) codes the current block Can be used to signal a prediction vector indicating the pixel position of the candidate block used for the purpose. The method 2600 ends at block 2620.

  [0184] In the method 2600, one or more of the blocks shown in FIG. 26 may be deleted (eg, not performed) and / or the order in which the methods are performed may be reversed. In some embodiments, additional blocks may be added to the method 2600. For example, in some embodiments, the coder currently blocks at least one pixel in the first line of the plurality of pixels in the current slice and at least one pixel in the third line of the plurality of pixels. Where a third line of pixels includes at least one pixel in the current block across the entire width of the current slice, where the third line is Different from line 1. Based on such determination, the coder determines (i) a cost associated with coding the current block based on the first block, where the first block is in the first region. At least one pixel in the second region and at least one pixel in the second region, and (ii) determining the current block based on a cost associated with coding the current block based on the first block A first block may be determined that is a candidate block used for prediction. In another embodiment, the coder includes the current block including at least one pixel in a first line of the plurality of pixels in the current slice and at least one pixel in a third line of the plurality of pixels. Where the third line of pixels includes at least one pixel in the current block across the entire width of the current slice, where the third line is the first line Is different. Based on such a determination, the coder determines (i) a first cost associated with coding the current block based on a first block having fewer pixels than the current block; Wherein the first block includes one or more pixels in each of the second regions, and (ii) the current block is based on a second block having the same number of pixels as the current block. Determining a second cost associated with coding, wherein the second block is one of each of the one or more pixels in the first block and each in the first region. And (iii) a candidate block used to predict the current block based on the determination that the second cost is greater than the first cost. It may determine a first block. Embodiments of the present disclosure are not limited to or by the example shown in FIG. 26, and other variations can be implemented without departing from the spirit of the present disclosure.

Other considerations
[0185] Information and signals disclosed herein may be represented using any of a variety of different technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that may be referred to throughout the above description are voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, light fields or optical particles, or any of them Can be represented by a combination.

  [0186] The various exemplary logic blocks and algorithm steps described with respect to the embodiments disclosed herein may be implemented as electronic hardware, computer software, or a combination of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Those skilled in the art may implement the described functionality in a variety of ways for a particular application, but such implementation decisions should not be construed as causing a departure from the scope of the present disclosure.

  [0187] The techniques described herein may be implemented in hardware, software, firmware, or any combination thereof. Such techniques may be implemented in any of a variety of devices, such as general purpose computers, wireless communication device handsets, or integrated circuit devices having multiple uses, including applications in wireless communication device handsets and other devices. Any feature described as a device or component may be implemented together in an integrated logical device or separately as a separate but interoperable logical device. When implemented in software, the techniques may be performed, at least in part, by a computer-readable data storage medium comprising program code that includes instructions that, when executed, perform one or more of the methods described above. Can be realized. The computer readable data storage medium may form part of a computer program product that may include packaging material. Computer readable media include random access memory (RAM) such as synchronous dynamic random access memory (SDRAM), read only memory (ROM), non-volatile random access memory (NVRAM), electrically erasable programmable read only memory (EEPROM) (Registered trademark)), flash memory, magnetic or optical data storage media, etc. The techniques may additionally or alternatively be carried by a computer readable communication medium that carries or communicates program code in the form of instructions or data structures, such as propagated signals or radio waves, and that can be accessed, read and / or executed by a computer. Can be realized at least in part.

  [0188] The program code may be one or more digital signal processors (DSPs), general purpose microprocessors, application specific integrated circuits (ASICs), field programmable logic arrays (FPGAs), or other equivalent integrated circuits or discrete logic. It may be executed by a processor that may include one or more processors, such as a circuit. Such a processor may be configured to perform any of the techniques described in this disclosure. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor is also implemented as a combination of computing devices, eg, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors associated with a DSP core, or any other such configuration. obtain. Accordingly, as used herein, the term “processor” refers to any of the above structures, any combination of the above structures, or any other structure or apparatus suitable for implementation of the techniques described herein. Can point to either. Further, in some aspects, the functionality described herein may be provided in dedicated software or hardware configured for encoding and decoding, or in a composite video encoder-decoder (codec). Can be incorporated. Also, the techniques may be fully implemented with one or more circuits or logic elements.

  [0189] The techniques of this disclosure may be implemented in a wide variety of devices or apparatuses, including a wireless handset, an integrated circuit (IC) or a set of ICs (eg, a chipset). In this disclosure, various components or units have been described to emphasize the functional aspects of a device configured to perform the disclosed techniques, but they necessarily require implementation by different hardware units. And not. Rather, as described above, the various units can be combined in a codec hardware unit, including one or more processors described above, or interoperable with appropriate software and / or firmware. It can be given by a set of operable hardware units.

  [0190] Although described above with respect to various different embodiments, features or elements from one embodiment may be combined with other embodiments without departing from the teachings of the present disclosure. However, the combination of features between the embodiments is not necessarily limited thereto. Various embodiments of the disclosure have been described. These and other embodiments are within the scope of the following claims.

[0190] Although described above with respect to various different embodiments, features or elements from one embodiment may be combined with other embodiments without departing from the teachings of the present disclosure. However, the combination of features between the embodiments is not necessarily limited thereto. Various embodiments of the disclosure have been described. These and other embodiments are within the scope of the following claims.
The invention described in the scope of claims at the beginning of the application of the present application will be added below.
[C1]
A method for coding a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme, the method comprising:
Determining a candidate block used to predict a current block in a current slice; and the candidate block is within a plurality of pixel locations, each corresponding to a reconstructed pixel in the current slice The range of pixel locations includes at least (i) a first region that includes one or more first pixel locations in a first line of pixels in the current slice; and The first line of pixels includes at least one pixel in the current block and spans the entire width of the current slice; (ii) in a second line of pixels in the current slice; A second region including one or more second pixel locations, wherein the second line of pixels is any pixel in the current block. Although not included, comprises, over the entire width of the current slice,
Determining a prediction vector indicative of a pixel position of the candidate block within the range of a plurality of pixel positions, wherein the pixel position of the candidate block is one of the first region or the second region; It is in,
Coding the current block in a simplified block prediction mode, at least in part through signaling the prediction vector;
A method comprising:
[C2]
The method of C1, wherein the first line of pixels and the second line of pixels comprise two adjacent raster scan lines of the current slice.
[C3]
The method of C1, wherein the current block is a 1x2 partition in a 2x8 block predicted in a simplified block prediction mode.
[C4]
The method of C1, wherein the current block is a 2x2 partition within a 2x8 block predicted in a simplified block prediction mode.
[C5]
The range of pixel locations further includes a third region comprising one or more third pixel locations in the second line of pixels, wherein the one or more third pixels The method of C1, wherein a position does not include any pixel position in the second line that is collocated with respect to a pixel position in the first line that is part of the current block.
[C6]
The method of C5, wherein the second region and the third region occupy the same raster scan line.
[C7]
The first region includes a first number of pixel locations, the second region includes a second number of pixel locations, and the third region includes a third number of pixel locations. The method of C5, wherein the first number is greater than the second number and greater than the third number.
[C8]
The method of C7, wherein the first, second, and third numbers are different from each other.
[C9]
Determining a cost associated with coding the current block based on each potential candidate block of a plurality of potential candidate blocks; and Each corresponding to one of the first and second pixel locations in a second region;
Identifying one of the plurality of potential candidate blocks in the first and second regions having the lowest cost as the candidate block;
The method of C1, further comprising:
[C10]
The number of bits required to uniquely identify each pixel location in the range of pixel locations is equal to the first number, the method comprising:
Determining that the current block is within a predetermined area in the current slice;
Signaling the prediction vector using less than the first number of bits;
The method of C1, further comprising:
[C11]
Determining that the current block includes at least one pixel in the first line of pixels in the current slice and at least one pixel in a third line of pixels; The third line of pixels includes at least one pixel in the current block and spans the entire width of the current slice, wherein the third line is the first line. Different,
Determining a cost associated with coding the current block based on a first block; and the first block includes at least one pixel in the first region and in the second region. And at least one pixel of
Determining the first block to be the candidate block used to predict the current block based on the cost associated with coding the current block based on the first block; And
The method of C1, further comprising:
[C12]
Determining that the current block includes at least one pixel in the first line of pixels in the current slice and at least one pixel in a third line of pixels; The third line of pixels includes at least one pixel in the current block and spans the entire width of the current slice, wherein the third line is the first line. Different,
Determining a first cost associated with coding the current block based on a first block having fewer pixels than the current block; and Including one or more pixels in each of the regions of
Determining a second cost associated with coding the current block based on a second block having the same number of pixels as the current block; and Including all of the one or more pixels in a block and one or more additional pixels in each of the first regions;
Determining the first block to be the candidate block used to predict the current block based on a determination that the second cost is greater than the first cost;
The method of C1, further comprising:
[C13]
An apparatus for coding a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme, the apparatus comprising:
A memory configured to store one or more reconstructed pixels of a current slice of video data;
One or more processors in communication with the memory, the one or more processors comprising:
Determining a candidate block used to predict a current block in the current slice; and the candidate block is within a plurality of pixel locations, each corresponding to a reconstructed pixel in the current slice. And the range of pixel locations is at least (i) a first region that includes one or more first pixel locations in a first line of pixels in the current slice; and The first line of pixels comprises at least one pixel in the current block and spans the entire width of the current slice; (ii) in the second line of pixels in the current slice A second region that includes one or more second pixel locations, wherein the second line of pixels includes any pixel in the current block. Although not included cell, over the entire width of the current slice comprises,
Determining a prediction vector indicative of a pixel position of the candidate block within the range of a plurality of pixel positions, wherein the pixel position of the candidate block is one of the first region or the second region; It is in,
Coding the current block in a simplified block prediction mode, at least in part through signaling the prediction vector;
An apparatus configured to do.
[C14]
The apparatus of C13, wherein the first line of pixels and the second line of pixels comprise two adjacent raster scan lines of the current slice.
[C15]
The apparatus of C13, wherein the current block is a 1 × 2 partition in a 2 × 8 block predicted in a simplified block prediction mode.
[C16]
The apparatus of C13, wherein the current block is a 2 × 2 partition in a 2 × 8 block predicted in a simplified block prediction mode.
[C17]
The range of pixel locations further includes a third region comprising one or more third pixel locations in the second line of pixels, wherein the one or more third pixels The apparatus of C13, wherein a position does not include any pixel position in the second line that is collocated with respect to a pixel position in the first line that is part of the current block.
[C18]
The apparatus according to C17, wherein the second area and the third area occupy the same raster scan line.
[C19]
The first region includes a first number of pixel locations, the second region includes a second number of pixel locations, and the third region includes a third number of pixel locations. The apparatus according to C17, wherein the first number is greater than the second number and greater than the third number.
[C20]
The apparatus according to C19, wherein the first, second, and third numbers are different from each other.
[C21]
The one or more processors are:
Determining a cost associated with coding the current block based on each potential candidate block of a plurality of potential candidate blocks; and Each corresponding to one of the first and second pixel locations in a second region;
Identifying one of the plurality of potential candidate blocks in the first and second regions having the lowest cost as the candidate block;
The apparatus of C13, further configured to perform:
[C22]
The number of bits required to uniquely identify each pixel location in the range of pixel locations is equal to a first number, and the one or more processors are:
Determining that the current block is within a predetermined area in the current slice;
Signaling the prediction vector using less than the first number of bits;
The apparatus of C13, further configured to perform:
[C23]
The one or more processors are:
Determining that the current block includes at least one pixel in the first line of pixels in the current slice and at least one pixel in a third line of pixels; The third line of pixels includes at least one pixel in the current block and spans the entire width of the current slice, wherein the third line is the first line. Different,
Determining a cost associated with coding the current block based on a first block; and the first block includes at least one pixel in the first region and in the second region. And at least one pixel of
Determining the first block to be the candidate block used to predict the current block based on the cost associated with coding the current block based on the first block; And
The apparatus of C13, further configured to perform:
[C24]
The one or more processors are:
Determining that the current block includes at least one pixel in the first line of pixels in the current slice and at least one pixel in a third line of pixels; The third line of pixels includes at least one pixel in the current block and spans the entire width of the current slice, wherein the third line is the first line. Different,
Determining a first cost associated with coding the current block based on a first block having fewer pixels than the current block; and Including one or more pixels in each of the regions of
Determining a second cost associated with coding the current block based on a second block having the same number of pixels as the current block; and Including all of the one or more pixels in a block and one or more additional pixels in each of the first regions;
Determining the first block to be the candidate block used to predict the current block based on a determination that the second cost is greater than the first cost;
The apparatus of C13, further configured to perform:
[C25]
Non-transitory physical computer storage comprising code configured to code a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme, the code being executed when To the device,
Determining a candidate block used to predict a current block in a current slice; and the candidate block is within a plurality of pixel locations, each corresponding to a reconstructed pixel in the current slice The range of pixel locations includes at least (i) a first region that includes one or more first pixel locations in a first line of pixels in the current slice; and The first line of pixels includes at least one pixel in the current block and spans the entire width of the current slice; (ii) in a second line of pixels in the current slice; A second region including one or more second pixel locations, wherein the second line of pixels is any pixel in the current block. Although not included, comprises, over the entire width of the current slice,
Determining a prediction vector indicative of a pixel position of the candidate block within the range of a plurality of pixel positions, wherein the pixel position of the candidate block is one of the first region or the second region; It is in,
Coding the current block in a simplified block prediction mode, at least in part through signaling the prediction vector;
Non-temporary physical computer storage.
[C26]
The range of pixel locations further includes a third region comprising one or more third pixel locations in the second line of pixels, wherein the one or more third pixels The non-transitory physical computer storage of C25, wherein a location does not include any pixel location in the second line that is collocated with respect to a pixel location in the first line that is part of the current block. .
[C27]
The first region includes a first number of pixel locations, the second region includes a second number of pixel locations, and the third region includes a third number of pixel locations. The non-transitory physical computer storage according to C26, wherein the first number is greater than the second number and greater than the third number.
[C28]
A video coding device configured to code a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme, the video coding device comprising:
Means for determining a candidate block used to predict a current block in a current slice; and the candidate block is within a plurality of pixel locations each corresponding to a reconstructed pixel in the current slice And the range of pixel locations includes at least (i) a first region that includes one or more first pixel locations in a first line of pixels in the current slice; Wherein the first line of pixels includes at least one pixel in the current block and spans the entire width of the current slice; (ii) a second line of pixels in the current slice A second region including one or more second pixel locations therein, wherein the second line of pixels is any of the current block Although not included Kuseru spans the entire width of the current slice comprises,
Means for determining a prediction vector indicative of a pixel position of the candidate block within the range of a plurality of pixel positions; and the pixel position of the candidate block is the first region or the second region In one,
Means for coding the current block in a simplified block prediction mode, at least in part through signaling the prediction vector;
A video coding device comprising:
[C29]
The range of pixel locations further includes a third region comprising one or more third pixel locations in the second line of pixels, wherein the one or more third pixels The video coding device of C28, wherein a position does not include any pixel position in the second line that is collocated with respect to a pixel position in the first line that is part of the current block.
[C30]
The first region includes a first number of pixel locations, the second region includes a second number of pixel locations, and the third region includes a third number of pixel locations. The video coding device of C29, wherein the first number is greater than the second number and greater than the third number.

Claims (30)

A method for coding a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme, the method comprising:
Determining a candidate block used to predict a current block in a current slice; and the candidate block is within a plurality of pixel locations, each corresponding to a reconstructed pixel in the current slice The range of pixel locations includes at least (i) a first region that includes one or more first pixel locations in a first line of pixels in the current slice; and The first line of pixels includes at least one pixel in the current block and spans the entire width of the current slice; (ii) in a second line of pixels in the current slice; A second region including one or more second pixel locations, wherein the second line of pixels is any pixel in the current block. Although not included, comprises, over the entire width of the current slice,
Determining a prediction vector indicative of a pixel position of the candidate block within the range of a plurality of pixel positions, wherein the pixel position of the candidate block is one of the first region or the second region; It is in,
Coding the current block in a simplified block prediction mode, at least in part through signaling the prediction vector.   The method of claim 1, wherein the first line of pixels and the second line of pixels comprise two adjacent raster scan lines of the current slice.   The method of claim 1, wherein the current block is a 1 × 2 partition within a 2 × 8 block predicted in a simplified block prediction mode.   The method of claim 1, wherein the current block is a 2 × 2 partition within a 2 × 8 block predicted in a simplified block prediction mode.   The range of pixel locations further includes a third region comprising one or more third pixel locations in the second line of pixels, wherein the one or more third pixels The method of claim 1, wherein a location does not include any pixel location in the second line that is collocated with respect to a pixel location in the first line that is part of the current block.   The method of claim 5, wherein the second region and the third region occupy the same raster scan line.   The first region includes a first number of pixel locations, the second region includes a second number of pixel locations, and the third region includes a third number of pixel locations. 6. The method of claim 5, wherein the first number is greater than the second number and greater than the third number.   The method of claim 7, wherein the first, second, and third numbers are different from each other. Determining a cost associated with coding the current block based on each potential candidate block of a plurality of potential candidate blocks; and Each corresponding to one of the first and second pixel locations in a second region;
The method of claim 1, further comprising: identifying one of the plurality of potential candidate blocks in the first and second regions having the lowest cost as the candidate block. The number of bits required to uniquely identify each pixel location in the range of pixel locations is equal to the first number, the method comprising:
Determining that the current block is within a predetermined region in the current slice;
The method of claim 1, further comprising signaling the prediction vector using less than the first number of bits. Determining that the current block includes at least one pixel in the first line of pixels in the current slice and at least one pixel in a third line of pixels; The third line of pixels includes at least one pixel in the current block and spans the entire width of the current slice, wherein the third line is the first line. Different,
Determining a cost associated with coding the current block based on a first block; and the first block includes at least one pixel in the first region and in the second region. And at least one pixel of
Determining the first block to be the candidate block used to predict the current block based on the cost associated with coding the current block based on the first block; The method of claim 1, further comprising: Determining that the current block includes at least one pixel in the first line of pixels in the current slice and at least one pixel in a third line of pixels; The third line of pixels includes at least one pixel in the current block and spans the entire width of the current slice, wherein the third line is the first line. Different,
Determining a first cost associated with coding the current block based on a first block having fewer pixels than the current block; and Including one or more pixels in each of the regions of
Determining a second cost associated with coding the current block based on a second block having the same number of pixels as the current block; and Including all of the one or more pixels in a block and one or more additional pixels in each of the first regions;
Determining the first block to be the candidate block used to predict the current block based on a determination that the second cost is greater than the first cost. The method of claim 1. An apparatus for coding a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme, the apparatus comprising:
A memory configured to store one or more reconstructed pixels of a current slice of video data;
One or more processors in communication with the memory, the one or more processors comprising:
Determining a candidate block used to predict a current block in the current slice; and the candidate block is within a plurality of pixel locations, each corresponding to a reconstructed pixel in the current slice. And the range of pixel locations is at least (i) a first region that includes one or more first pixel locations in a first line of pixels in the current slice; and The first line of pixels comprises at least one pixel in the current block and spans the entire width of the current slice; (ii) in the second line of pixels in the current slice A second region that includes one or more second pixel locations, wherein the second line of pixels includes any pixel in the current block. Although not included cell, over the entire width of the current slice comprises,
Determining a prediction vector indicative of a pixel position of the candidate block within the range of a plurality of pixel positions, wherein the pixel position of the candidate block is one of the first region or the second region; It is in,
An apparatus configured to code the current block in a simplified block prediction mode, at least in part through signaling the prediction vector.   14. The apparatus of claim 13, wherein the first line of pixels and the second line of pixels comprise two adjacent raster scan lines of the current slice.   The apparatus of claim 13, wherein the current block is a 1 × 2 partition within a 2 × 8 block predicted in a simplified block prediction mode.   The apparatus of claim 13, wherein the current block is a 2 × 2 partition within a 2 × 8 block predicted in a simplified block prediction mode.   The range of pixel locations further includes a third region comprising one or more third pixel locations in the second line of pixels, wherein the one or more third pixels 14. The apparatus of claim 13, wherein a location does not include any pixel location in the second line that is collocated with respect to a pixel location in the first line that is part of the current block.   The apparatus of claim 17, wherein the second region and the third region occupy the same raster scan line.   The first region includes a first number of pixel locations, the second region includes a second number of pixel locations, and the third region includes a third number of pixel locations. The apparatus of claim 17, wherein the first number is greater than the second number and greater than the third number.   The apparatus of claim 19, wherein the first, second, and third numbers are different from each other. The one or more processors are:
Determining a cost associated with coding the current block based on each potential candidate block of a plurality of potential candidate blocks; and Each corresponding to one of the first and second pixel locations in a second region;
The method of claim 13, further comprising: identifying one of the plurality of potential candidate blocks in the first and second regions having the lowest cost as the candidate block. The device described. The number of bits required to uniquely identify each pixel location in the range of pixel locations is equal to a first number, and the one or more processors are:
Determining that the current block is within a predetermined region in the current slice;
14. The apparatus of claim 13, further configured to: signal the prediction vector using less than the first number of bits. The one or more processors are:
Determining that the current block includes at least one pixel in the first line of pixels in the current slice and at least one pixel in a third line of pixels; The third line of pixels includes at least one pixel in the current block and spans the entire width of the current slice, wherein the third line is the first line. Different,
Determining a cost associated with coding the current block based on a first block; and the first block includes at least one pixel in the first region and in the second region. And at least one pixel of
Determining the first block to be the candidate block used to predict the current block based on the cost associated with coding the current block based on the first block; 14. The apparatus of claim 13, further configured to: The one or more processors are:
Determining that the current block includes at least one pixel in the first line of pixels in the current slice and at least one pixel in a third line of pixels; The third line of pixels includes at least one pixel in the current block and spans the entire width of the current slice, wherein the third line is the first line. Different,
Determining a first cost associated with coding the current block based on a first block having fewer pixels than the current block; and Including one or more pixels in each of the regions of
Determining a second cost associated with coding the current block based on a second block having the same number of pixels as the current block; and Including all of the one or more pixels in a block and one or more additional pixels in each of the first regions;
Determining the first block to be the candidate block used to predict the current block based on a determination that the second cost is greater than the first cost. The apparatus of claim 13 further configured. Non-transitory physical computer storage comprising code configured to code a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme, the code being executed when To the device,
Determining a candidate block used to predict a current block in a current slice; and the candidate block is within a plurality of pixel locations, each corresponding to a reconstructed pixel in the current slice The range of pixel locations includes at least (i) a first region that includes one or more first pixel locations in a first line of pixels in the current slice; and The first line of pixels includes at least one pixel in the current block and spans the entire width of the current slice; (ii) in a second line of pixels in the current slice; A second region including one or more second pixel locations, wherein the second line of pixels is any pixel in the current block. Although not included, comprises, over the entire width of the current slice,
Determining a prediction vector indicative of a pixel position of the candidate block within the range of a plurality of pixel positions, wherein the pixel position of the candidate block is one of the first region or the second region; It is in,
Non-transitory physical computer storage that causes the current block to be coded in a simplified block prediction mode, at least in part through signaling the prediction vector.   The range of pixel locations further includes a third region comprising one or more third pixel locations in the second line of pixels, wherein the one or more third pixels 26. The non-transitory physical of claim 25, wherein a location does not include any pixel location in the second line that is collocated with respect to a pixel location in the first line that is part of the current block. Computer storage.   The first region includes a first number of pixel locations, the second region includes a second number of pixel locations, and the third region includes a third number of pixel locations. 27. The non-transitory physical computer storage of claim 26, wherein the first number is greater than the second number and greater than the third number. A video coding device configured to code a block of video data in a simplified block prediction mode of a constant bit rate video coding scheme, the video coding device comprising:
Means for determining a candidate block used to predict a current block in a current slice; and the candidate block is within a plurality of pixel locations each corresponding to a reconstructed pixel in the current slice And the range of pixel locations includes at least (i) a first region that includes one or more first pixel locations in a first line of pixels in the current slice; Wherein the first line of pixels includes at least one pixel in the current block and spans the entire width of the current slice; (ii) a second line of pixels in the current slice A second region including one or more second pixel locations therein, wherein the second line of pixels is any of the current block Although not included Kuseru spans the entire width of the current slice comprises,
Means for determining a prediction vector indicative of a pixel position of the candidate block within the range of a plurality of pixel positions; and the pixel position of the candidate block is the first region or the second region In one,
Means for coding the current block in a simplified block prediction mode, at least in part through signaling the prediction vector.   The range of pixel locations further includes a third region comprising one or more third pixel locations in the second line of pixels, wherein the one or more third pixels 30. The video coding device of claim 28, wherein a location does not include any pixel location in the second line that is collocated with respect to a pixel location in the first line that is part of the current block.   The first region includes a first number of pixel locations, the second region includes a second number of pixel locations, and the third region includes a third number of pixel locations. 30. The video coding device of claim 29, wherein the first number is greater than the second number and greater than the third number.

Images